GSA Bulletin; January 2008; v. 120; no. 1-2;
p. 54-84; DOI: 10.1130/B26073.1
© 2008 Geological Society of America
Synoptic reconstruction of a major ancient lake system: Eocene Green River Formation, western United States
M. Elliot Smith*,1,
Alan R. Carroll1 and
Brad S. Singer1
1 Department of Geology and Geophysics, University of Wisconsin, 1215 West Dayton Street, Madison, Wisconsin 53706, USA
 |
ABSTRACT
|
|---|
Numerous 40Ar/39Ar experiments on sanidine and biotite from 22 ash beds and 3 volcaniclastic sand beds from the Greater Green River, Piceance Creek, and Uinta Basins of Wyoming, Colorado, and Utah constrain 
8 m.y. of the Eocene Epoch. Multiple analyses were conducted per sample using laser fusion and incremental heating techniques to differentiate inheritance, 40Ar loss, and 39Ar recoil. When considered in conjunction with existing radioisotopic ages and lithostratigraphy, biostratigraphy, and magnetostratigraphy, these new age determinations facilitate temporal correlation of linked Eocene lake basins in the Laramide Rocky Mountain region at a significantly increased level of precision. To compare our results to the geomagnetic polarity time scale and the regional volcanic record, the ages of Eocene magnetic anomalies C24 through C20 were recalibrated using seven 40Ar/39Ar ages. Overall, the ages obtained for this study are consistent with the isochroneity of North American land-mammal ages throughout the study area, and provide precise radioisotopic constraints on several important biostratigraphic boundaries.
Applying these new ages, average sediment accumulation rates in the Greater Green River Basin, Wyoming, were approximately three times faster at the center of the basin versus its ramp-like northern margin during deposition of the underfilled Wilkins Peak Member. In contrast, sediment accumulation occurred faster at the edge of the basin during deposition of the balanced filled to overfilled Tipton and Laney Members. Sediment accumulation patterns thus reflect basin-center–focused accumulation rates when the basin was underfilled, and supply-limited accumulation when the basin was balanced filled to overfilled. Sediment accumulation in the Uinta Basin, at Indian Canyon, Utah, was relatively constant at 150 mm/k.y. during deposition of over 5 m.y. of both evaporative and fluctuating profundal facies, which likely reflects the basin-margin position of the measured section. The most rapid sediment accumulation for the entire system (>1 m/k.y.) occurred between 49.0 and 47.5 Ma, when volcaniclastic materials from the Absaroka and/or Challis volcanic fields entered the Green River Formation lakes from the north.
Our new ages combined with existing paleomagnetic and biostratigraphic control permit the first detailed synoptic comparison of lacustrine depositional environments in all the Green River Formation basins. Coupled with previously published paleocurrent observations, our detailed correlations show that relatively freshwater lakes commonly drained into more saline downstream lakes. The overall character of Eocene lake deposits was therefore governed in part by the geomorphic evolution of drainage patterns in the surrounding Laramide landscape. Freshwater (overfilled) lakes were initially dominant (53.5–52.0 Ma), possibly related to high erosion rates of remnant Cretaceous strata on adjacent uplifts. Expansion of balanced-fill lakes first occurred in all Green River Formation basins at 52.0–51.3 Ma and again between 49.6 and 48.5 Ma. Evaporative (underfilled) lakes occurred in various basins between 51.3 and 45.1 Ma, coincident with the end of the early Eocene climatic optima and subsequent onset of global cooling defined from marine record. However, evaporite intervals in the different depocenters were deposited at different times rather than being confined to a single episode of arid climate. Evaporative terminal sinks were initially located in the Greater Green River and Piceance Creek Basins (51.3–48.9 Ma), then gradually migrated southward to the Uinta Basin (47.1–45.2 Ma). This history is likely related to progressive southward construction of the Absaroka Volcanic Province, which constituted a major topographic and thermal anomaly that contributed to a regional north to south hydrologic gradient. The Greater Green River and Piceance Creek Basins were eventually filled from north to south with Absarokaderived detritus at sedimentation rates 1–2 orders of magnitude greater than the underlying lake deposits.
Key Words: Ar-Ar • Absaroka • Uinta Basin • Piceance Creek Basin • land-mammal ages • lake type • Laramide • Eocene • Green River Formation
|
INTRODUCTION
|
|---|
Large lakes are widely recognized for their importance as economic resources and as archives of faunal, floral, and climatic evolution (e.g., Bradley, 1929; Franczyk et al., 1989; Wilf, 2000) but are less well understood with regard to the geomorphic evolution of the landscapes surrounding them (cf. Surdam and Stanley, 1980; Pietras et al., 2003a; Carroll et al., 2006). The Eocene Green River Formation (Hayden, 1869) of Wyoming, Colorado, and Utah represents one of the best-documented ancient lake systems and has long been a type example for understanding lacustrine depositional systems (Bradley, 1929; Eugster and Surdam, 1973; Carroll and Bohacs, 1999). Since Marsh (1871) speculated that the Green River Formation lakes were hydrologically connected, numerous authors have proposed temporal correlations of its strata across the Uinta uplift (e.g., Bradley, 1931; Roehler, 1974; Surdam and Stanley, 1980). However, due to the absence of intervening strata between depocenters (Fig. 1), lateral facies changes, and limited radioisotopic age control, temporal correlation has been insufficiently precise to address more specific questions concerning basin evolution. This paper provides an 40Ar/39Ar–based age framework that allows for the first detailed delineation on upstream-downstream relationships between the sequences of lakes that occupied the Green River Formation basins and provides a fundamental measurement of lacustrine sediment accumulation rates over 8 m.y. of the Eocene Epoch.
Until recently, mammalian biostratigraphy was the only available method for determining the relative age of strata in the terrestrial basins that contain the Green River Formation (Wood et al., 1941; Lillegraven, 1993; Robinson et al., 2004). However, mammalian fossils are typically preserved in adjacent alluvial deposits that can be difficult to correlate with lake deposits using the physical characteristics of the strata (cf. Clyde et al., 2004; Smith et al., 2004). The temporal resolution of mammalian biostratigraphy is also fundamentally limited by the richness of collections at particular sites. Moreover, phenomena such as diachronous first and last occurrences of taxa caused by taphonomic biases (Smith and Holroyd, 2003) and geographic or climatic heterogeneity (Gunnell and Bartels, 2001) can only be assessed using geochronology that is independent of mammalian biostratigraphy.
Paleomagnetic polarity records have also been used to establish the timing of Eocene terrestrial strata and have the potential to provide precise relative age control (e.g., Tauxe et al., 1994; Clyde et al., 2001). However, paleomagnetic records are inherently binary and can be hampered by changing sedimentation rates, lacuna, poor remanence acquisition, and magnetic overprinting. Comparisons between magnetic polarity records from continental strata and the geomagnetic polarity time scale (Cande and Kent, 1992, 1995; Ogg and Smith, 2004) have proven problematic because higher numbers of polarity reversals are often preserved in terrestrial sediments than are recorded by ocean-floor magnetic anomalies (Elston et al., 1994). Nevertheless, paleomagnetic stratigraphy provides a vital component of terrestrial geochronology but also requires adequate independent calibration.
Several K-Ar and early 40Ar/39Ar geochronology efforts were undertaken in the Green River Formation and related Eocene continental strata prior to ca. 1990 using phenocrysts from ash beds (Evernden et al., 1964; Mauger, 1977; O'Neill, 1980). However, due to the lower sensitivity of older instruments and difficulty in acquiring sufficient quantities of sanidine, these studies focused on large aliquots of biotite, which tend to be more susceptible to chemical weathering than sanidine (Renne, 2000; Smith et al., 2006). Age determinations were further limited in accuracy by the unavoidable inclusion of inherited or altered biotite grains in the large samples analyzed (cf. Smith et al., 2006), and limited in their precision by lower resolution mass spectrometry. Recently, 40Ar/39Ar geochronologic studies of tuff beds using smaller samples and more sensitive mass spectrometry have begun to significantly refine the timing of the Green River Formation and related alluvial strata (Wing et al., 1991; Smith et al., 2003, 2004, 2006). This study integrates 25 new age determinations with detailed facies and geochemical analyses to construct the most comprehensive and highly resolved chronostratigraphic model available for any major pre-Quaternary lake system.
|
GEOLOGIC SETTING
|
|---|
The Green River Formation of Wyoming, Colorado, and Utah was deposited in a series of continental basins that occupy a broken foreland province to the east of the Cordilleran fold and thrust belt. These basins are separated from one another by chains of anticlinal basement-cored uplifts that collectively comprise the Laramide orogeny, and were variably active from the Cretaceous through Eocene (Fig. 1; Beck et al., 1988; Dickinson et al., 1988). The formation has a maximum thickness of nearly 2 km and spans much of the early and middle Eocene Epoch (Figs. 2
and 3). The Green River Formation basins are part of a suite of basins that have been differentiated based on their structural setting and strata into four principle types: ponded, perimeter, axial, and extensional (Dickinson et al., 1988; Constenius, 1996; Fig. 4; see GSA Data Repository Table DR11). Ponded basins, for which the Green River Formation basins are the type example, are bounded by basement-involved uplifts and contain evidence for internal drainage during at least portion of their history. They typically contain thick packages (3–5 km) of alluvial and lacustrine strata (Dickinson et al., 1988; Baars et al., 1988). Perimeter basins occur on the east edge of the broken foreland province and contain alluvial strata with east-directed paleocurrent indicators that indicate external drainage (Dickinson et al., 1988). Axial basins are small, elongate, intermontane basins that formed amid the main series of basement uplifts of central Colorado and southeastern Wyoming (Fig. 1). These basins typically contain 1–2 km packages of coarsegrained alluvial strata (Dickinson et al., 1988). Extensional basins are strike-elongate grabens and half grabens that overlie the former Cordilleran fold and thrust belt, are commonly bounded by normal faults that reactivate Cordilleran thrust faults, and often contain thick (2–5 km) but areally restricted packages of alluvial and lacustrine strata (Constenius, 1996).
View larger version (61K):
[in this window]
[in a new window]
|
Figure 2. Lithostratigraphic and time stratigraphic cross sections of Eocene strata in the Greater Green River, Piceance Creek, and Uinta Basins along cross-section X-X' (see Fig. 1) showing the stratigraphic position of facies associations, structural features, and dated tuff beds. Cross-section line was chosen in order to intersect area of thickest sediment accumulation, sites of bedded evaporites, and principle sills. Inset columns with white background depict stratigraphy and chronostratigraphy of the Green River Formation in the Fossil Basin and Wasatch Plateau region. The stratigraphic references for numbered segments used to construct the cross section are in Table DR7 (see footnote 1). (Continued on following page).
|
|
View larger version (35K):
[in this window]
[in a new window]
|
Figure 3. Composite age model for the Green River Formation and associated strata in the Greater Green River and Uinta–Piceance Creek Basins showing correlation of radioisotopic ages, biostratigraphy, and magnetostratigraphy. Lithostratigraphic symbols as in Figure 2 . A complete list of biostratigraphic references for North American land-mammal ages is in Table DR1 (see footnote 1) (cf. Robinson et al., 2004). All radioisotopic ages have been normalized to the intercalibration values of Renne et al. (1998) and are shown with 2 analytical and intercalibration uncertainties. GPTS—geomagnetic polarity time scale, recalibrated from Cande and Kent (1992, 1995) to modern intercalibration values (see Fig. 9; Tables 4 and 5) (cf. Renne et al., 1998; Smith et al., 2003). Columns labeled P1a, P1b, P2, and P3 illustrate composite magnetostratigraphy for the western and eastern Uinta, Washakie, and Bridger Basins, respectively (P1a—Prothero, 1996; P1b—Prothero, 1996; P2—Stucky et al., 1996; McCarroll et al., 1996a; P3—Jerskey, 1981; Clyde et al., 1997, 2001). Letters A, B1, B2, and C adjacent to Uinta Basin column refer to divisions of the Uinta Formation (O.A. Peterson in Osborn, 1895, p. 72–74; cf. Prothero, 1996). Numerical subdivisions of the Adobe Town Member of the Washakie Formation in the Washakie Basin are from McCarroll et al. (1996a). Uppercase letters adjacent to Bridger Basin column refer to Bridger Beds (Matthew, 1909; cf. Evanoff et al., 1998). Regional tectonic, volcanic, and paleofloral records and global paleoclimatologic data from benthic forams are all recalibrated to the standard ages of Renne et al. (1998). All paleofloral-based precipitation estimates and sites are from Wilf (2000), except the Bonanza site in the Uinta Basin (P. Wilf and K. Johnson, 2006, personal commun.). Sample information for Bonanza site: from 28 dicot species, 894 specimens, in the Bonanza flora excavated by K. Johnson (Denver Museum of Natural History locations 323 and 1732; see Wilf et al., 2001): %Leptophyll = 0.012, %Nanophyll = 0.223, %Microphyll = 0.539, %Notophyll = 0.187, %Mesophyll = 0.039; mean ln (leaf area, mm2) = 6.35.
|
|
Volcanism occurred over broad areas of the northwestern United States during the Eocene and provided both fallout tuffs and volcaniclastic sediment to the Green River Formation lake basins (Fig. 1; Surdam and Stanley, 1980; Fritz and Harrison, 1985; Armstrong and Ward, 1991). Major volcanic centers include the Absaroka Volcanic Province, Challis volcanic field, and Lowland Creek Volcanics; minor fields are scattered throughout the region (Fig. 1). The stratigraphic and time-stratigraphic constraints for Eocene volcanic fields in Wyoming, Montana, and Idaho are summarized in Figure 5.
View larger version (68K):
[in this window]
[in a new window]
|
Figure 5. Map showing major eruptive centers and lithostratigraphic and time-stratigraphic cross section of volcanic and volcaniclastic Eocene strata in the Absaroka Volcanic Province, compiled from 29 stratigraphic studies. Mapping modified from Smedes and Prostka (1972). A complete list of lithostratigraphic, biostratigraphic, and radioisotopic references for numbered sites is in Table DR8 (see footnote 1). Note that strata are portrayed at the same scale as in Figure 2. 40Ar/39Ar ages are from Hiza (1999) except: (a) Feeley and Cosca (2003), (b) Harlan et al. (1996), (c) Wilson and Elliott (1997), (d) Ispolatov (1997), (e) Snider (1995), (f) Snider and Moye (1989), (g) Janecke et al. (1997), (h) Janecke and Snee (1993), (i) M'Gonigle and Dalrymple (1996), (j) O'Neill et al. (2004), (k) Wallace et al. (1992, K/Ar), and (l) House et al. (2002). All 40Ar/39Ar ages are shown with 2 intercalibration uncertainties relative to the standard values of Renne et al. (1998). Note that the chronostratigraphic diagram does not imply internally uniform rates of accumulation, and significant interstratal lacunae are undoubtedly present, particularly within more proximal volcanic deposits.
|
|
|
NONMARINE SEDIMENTARY FACIES
|
|---|
Lake Types
Representing a broad range of lacustrine facies, several thick lenses of the Green River Formation occupy two principle basins: the Greater Green River Basin and the Uinta–Piceance Creek Basin, which are separated from one another by the east-west–trending, anticlinal Uinta uplift (Fig. 1; Bradley, 1964; Johnson, 1985; Roehler, 1992a). The names Lake Gosiute (King, 1878, p. 446) and Lake Uinta (Bradley, 1931) were assigned to the lakes that existed in the northern (Greater Green River) and southern (Uinta–Piceance Creek) basins, respectively. Each of these lakes varied greatly in their chemistries and areal extents during the course of Green River Formation deposition (Fig. 2A).
Distinctive assemblages of lithologies and fossils within the Green River Formation allow for its subdivision into three principle lacustrine facies associations: fluvial lacustrine, fluctuating profundal, and evaporative (Carroll and Bohacs, 1999). These associations are the basis for interpreting lake type, which reflects the long-term balance between potential accommodation and water plus sediment fill (Carroll and Bohacs, 1999; Bohacs et al., 2000; Carroll and Bohacs, 2001). Carroll and Bohacs (1999) defined the facies associations and lake type interpretations for the Green River Formation in the Greater Green River Basin. In this study we have extended these interpretations to Green River Formation strata in the Uinta, Piceance Creek, and Fossil Basins (Fig. 1). Although a wide variety of characteristics has been utilized to define lacustrine facies associations (Horsfield et al., 1994; Bohacs et al., 2000), several key features outlined here and in Table 1 provide the strongest evidence for these associations. Evaporative facies are best recognized via the presence of bedded evaporites and absence of fish fossils, and are interpreted to represent the deposits of hypersaline lakes within underfilled basins in which water rarely rose above the level of the downstream sill. Fluviallacustrine facies preserve abundant mollusc fossils and occasional fish fossils, and are interpreted to have been deposited from freshwater lakes in overfilled basins where water consistently spilled over the downstream sill. Fluctuating profundal facies are typically composed of laminated, organic-rich carbonate mudstones intercalated with thin desiccation horizons, and are interpreted to represent the deposits of brackish to saline lakes that occupied balanced-filled basins where water oscillated near the sill level. Assignments of facies association and lake type in most cases correspond to previously identified stratal units and are primarily employed to help standardize terminology between the basins.
Alluvial Facies
Alluvial strata surround and interfinger with the Green River Formation (Fig. 2A; Roehler, 1992a). For this study, alluvial deposits have been subdivided into three broad facies associations according to their mode of deposition: deltaic, alluvial plain, and alluvial fan. Deltaic deposits signify influxes of water and sediment from rivers and typically consist of well-sorted, coarsening- and shallowing-upward packages of sandstone and siltstone that exhibit progradational stratal geometries (Fig. 2A). Alluvial plain facies are typically composed of mud, silt, and sand deposits that are often channelized and pedogenically altered, reflecting the avulsion of streams over broad, exposed plains (Braunagel and Stanley, 1977; Roehler, 1993). Alluvial fan deposits typically consist of moderately to poorly sorted sand- to boulder-sized clasts derived from and deposited adjacent to basin-bounding uplifts. They signify the downstream termini of short, steep drainage networks and record the denudation of the uplifts from which the sediments were eroded (Crews and Ethridge, 1993; Carroll et al., 2006). Alluvial deposits as a whole have also been differentiated into two petrographic types: (1) locally derived feldspathic, sublithic, and quartz arenite sandstones; and (2) volcaniclithic deposits delivered from contemporaneous volcanic fields (Fig. 2A; Surdam and Stanley, 1980; Dickinson et al., 1986).
|
REGIONAL LITHOSTRATIGRAPHY
|
|---|
Greater Green River Basin
The Greater Green River Basin consists of four subbasins (Bridger, Great Divide, Washakie, and Sand Wash) that are partitioned from one another by the north-south–trending Rock Springs uplift and several smaller east-west–trending structures (Fig. 1; Love et al., 1961). Each subbasin contains a unique succession of strata, but all record a long-term evolution from open to closed and return to open hydrologic conditions during deposition of the Green River Formation (Fig. 2A; Roehler, 1993; Carroll and Bohacs, 1999). The Luman, Tipton, and Wilkins Peak Members record a progression from fluviallacustrine to fluctuating profundal through evaporative facies. The evaporative Wilkins Peak Member is primarily restricted to the Bridger sub-basin and contains bedded evaporites, predominantly trona, shortite, and halite (Fahey, 1962; Pietras et al., 2003a), and is laterally equivalent to alluvial deposits of the Cathedral Bluffs Member of the Wasatch Formation in adjacent subbasins (Sullivan, 1985; Roehler, 1992a). The fluctuating profundal to fluviallacustrine LaClede bed of the Laney Member overlies the Wilkins Peak Member and records an expansion of lake facies into all of the subbasins of the Greater Green River Basin (Fig. 2A; Surdam and Stanley, 1979; Roehler, 1992a). Volcaniclastic alluvium of the Sand Butte bed of the Laney Member, Bridger Formation, and lower Washakie Formation replace lacustrine strata from north to south in a time-transgressive fashion (Stanley and Surdam, 1978; McCarroll et al., 1996a; Evanoff et al., 1998).
Fossil Basin
The smaller, wedge-top Fossil Basin occupies the fold and thrust belt to the west of the Greater Green River Basin (Fig. 1; Oriel and Tracey, 1970; Lamerson, 1982; DeCelles and Currie, 1996; Chandler, 2006). The fluviallacustrine to fluctuating profundal Fossil Butte Member overlies the alluvial Wasatch Formation and is overlain by the evaporative Angelo Member (Fig. 2A). The alluvial Bullpen Member of the Wasatch Formation overlies the Angelo Member (Fig. 2A; Oriel and Tracey, 1970; Buchheim, 1994; Buchheim and Eugster, 1998). In the Fowkes and Woodruff Basins to the west of the Fossil Basin, the Bullpen Member is unconformably overlain by volcaniclastic alluvial strata of the Fowkes Formation (Figs. 1 and 2A; Oriel and Tracey, 1970; Nelson, 1973).
Piceance Creek Basin
As in the Greater Green River Basin, lacustrine strata in the Piceance Creek Basin record a progression from open to closed and return to open hydrologic conditions (Fig. 2A). Alluvial deposits in the Piceance Creek and Uinta Basins are physically separated by the Douglas Creek arch, and from strata in the Greater Green River Basin by the Uinta uplift and Axial Basin arch (Figs. 1 and 2). To more clearly differentiate between these strata, we have adopted the names DeBeque Formation (Piceance Creek Basin) and Colton Formation (Uinta Basin) in place of Wasatch Formation for alluvial deposits underlying and interfingering with the Green River Formation in these basins (cf. Powell, 1876; Bradley, 1964; Robinson et al., 2004).
In the center of the Piceance Creek Basin, the mollusc-bearing Cow Ridge Member of the Green River Formation (Johnson, 1984) overlies the alluvial DeBeque Formation (Donnell, 1961b; Kihm, 1984) and is overlain by the Garden Gulch Member of the Green River Formation, which lacks molluscs (Bradley, 1931; Johnson, 1985). Above the Garden Gulch Member, the evaporative lower Parachute Creek Member (Donnell, 1961a; Trudell et al., 1974) is intercalated with both bedded and disseminated evaporites, predominantly nacholite and halite (Bradley, 1931; Dyni, 1981).
Lacustrine strata overlying the eastern flank of the Douglas Creek arch have been collectively referred to the Douglas Creek Member (Bradley, 1931; Donnell, 1961a). However, several features of these deposits suggest genetic ties with their lateral equivalents in the basin center. At Douglas Pass, which overlies the Douglas Creek arch, lacustrine strata equivalent to the Cow Ridge and Garden Gulch Members contain gastropods at their base but none above, mirroring the shift to fluctuating profundal facies observed in the basin center (Moncure and Surdam, 1980; Johnson et al., 1988). Strata equivalent to the lower Parachute Creek Member on the Douglas Creek arch contain evaporite casts and abundant exposure horizons, consistent with the evaporative facies in the basin center (Moncure and Surdam, 1980; Cole, 1985).
The upper part of the Parachute Creek Member contains the Mahogany zone (Bradley, 1931; Cashion, 1967), a 20–60 m interval predominantly composed of laminated organicrich micrite that extends across the Douglas Creek arch into both the Uinta and Piceance Creek Basins (Fig. 2A; Cashion and Donnell, 1972; Remy, 1992). Although it represents the broadest expansion of Lake Uinta, the Mahogany zone in the Piceance Creek Basin contains evaporites (Trudell et al., 1973; Dyni, 1981) and largely lacks fish fossils in the Uinta Basin (Cashion, 1967; Remy, 1992). In addition, a tuff bed within the Mahogany zone in the Piceance Creek Basin exhibits K-spar alteration of its formerly glassy ash matrix, denoting deposition in alkaline lake water with an elevated solute concentration (Surdam and Parker, 1972; Mason, 1983). We have therefore categorized the facies of the Mahogany zone as evaporative rather than fluctuating profundal. Above the Mahogany zone in the Piceance Creek Basin, the Parachute Creek Member is composed of fluctuating profundal facies and interfingers with the volcaniclastic deltaic and alluvial Uinta Formation (Trudell et al., 1970; Hail, 1987).
Uinta Basin
The Green River Formation achieves its greatest thickness in the Uinta Basin and records an open to closed to open hydrologic trajectory similar to that observed in the other two major basins (Fig. 2A). Unfortunately, stratigraphic terminology in the Uinta Basin is beset by informal and overlapping designations, due in part to limited surface exposure (cf. Remy, 1992). In the interest of consistency, we have adopted the informal stratigraphic designations of Weiss et al. (1990) and Remy (1992) but recommend future adoption of nongenetic formal stratal designations for these units. At the base of the Eocene succession, nearly 1000 m of lacustrine deposits occupy the subsurface depocenter of the basin (Ryder et al., 1976; Fouch, 1981) and are equivalent to mollusc-bearing fluviallacustrine facies exposed at the basin margins (Bradley, 1931; Cashion, 1967; Remy, 1992). This interval is also correlative to the upper part of the alluvial Colton Formation (Spieker, 1946; Cashion, 1967; Pusca, 2003). These basal fluviallacustrine strata are overlain by 200 m of fluctuating profundal strata assigned to the transitional interval (Remy, 1992; Pusca, 2003). The evaporative upper member overlies the transitional interval and contains the Mahogany zone at its base (Bradley, 1931; Cashion, 1967; Remy, 1992). In the eastern Uinta Basin, the upper member interfingers with and is overlain by the alluvial Uinta Formation (Douglass, 1914; Cashion, 1967). However, in the western Uinta Basin, the Green River Formation above the upper member thickens to >500 m. Two additional units are preserved as a result: the evaporative, evaporite-bearing saline facies, and the overlying fluctuating profundal to fluviallacustrine sandstone and limestone facies above (Dane, 1955; Dyni et al., 1985; Weiss et al., 1990). These units are equivalent to the lower part of the alluvial Uinta Formation in the eastern Uinta Basin, and are overlain by the upper Uinta Formation (Fig. 2A; Dane, 1955; Prothero, 1996).
High Plateaus of Utah
Extending southward from the Uinta Basin into central Utah along the margin of the Sevier fold and thrust belt are isolated exposures of Green River Formation (Figs. 1 and 2; Spieker, 1949; Doelling, 1972). These strata overlie the alluvial Colton Formation and consist of evaporative to fluviallacustrine strata that are loosely correlated to the saline facies and sandstone and limestone facies (McGookey, 1960; Sheliga, 1980). Overlying the Green River Formation in the eastern half of the region is the alluvial Crazy Hollow Formation (Weiss and Warner, 2001), whereas the volcaniclastic alluvial Golden's Ranch Formation (Muessig, 1951; Doelling, 1972) is the uppermost Eocene unit in the western half. Farther to the southwest along the fold and thrust belt, Eocene alluvial and lacustrine strata are referred to as the Claron Formation (Figs. 1 and 4; Goldstrand, 1994).
|
GEOCHRONOLOGY
|
|---|
40Ar/39Ar Methodology
The geochronology of Green River Formation and associated strata was accomplished via 40Ar/39Ar dating of volcanic phenocrysts preserved in ash beds and volcaniclastic sandstone beds (cf. Smith, 2007). The majority of dated units are preserved within laminated to finely bedded lacustrine facies (Table DR2 [see footnote 1]; cf. Smith et al., 2003). The origins of the names of sampled units are explained in Table DR2. Samples were collected from the base of tuff beds, which are often subtly graded, in order to maximize phenocryst grain size and limit contamination by admixed detrital grains. Minerals for dating were obtained using the separation techniques outlined in Smith et al. (2003, 2006) and were irradiated together with flux monitors at the Oregon State University Triga reactor (details of J value calculations are found in Data Repository Fig. DR1). Ar isotopic compositions were determined using a CO2 laser to fuse or incrementally heat sanidine or biotite crystals following the procedures detailed in Smith et al. (2003, 2006).
Ages were determined on the basis of 2234 analyses of phenocrysts from 25 tuffaceous samples (Figs. 6–8; Tables 2 and DR3). Age plateaus are here defined as three or more contiguous, concordant steps containing at least 50% of the total 39Ar released. Heating steps were considered concordant if the mean squared weighted deviation (MSWD) resulting from their inclusion was less than the Students-T distribution limit for the number of included steps (Koppers, 2002). When MSWD exceeded 1, analytical errors were multiplied by the square root of the MSWD (cf. York, 1969; Koppers, 2002). Plateau ages are the weighted mean of included steps, whereas integrated (total fusion) ages combine the Ar released during all heating steps. Inverse-variance weighted mean ages and uncertainties were calculated from both fusion and plateau ages according to Taylor (1982) using Isoplot 3.00 (Ludwig, 2003). An arbitrary outlier exclusion criteria adapted from Deino and Potts (1990) was applied, in which apparent plateau and fusion ages were excluded if they contributed to a MSWD >1.5, thereby eliminating only obvious outliers from the age calculation (Smith et al., 2006). Isochrons were regressed using the method of York (1969) in order to test for excess argon, and in virtually all cases exhibit atmospheric intercepts (Fig. DR2; see footnote 1).
View larger version (26K):
[in this window]
[in a new window]
|
Figure 6. Cumulative probability diagrams summarizing 40Ar/39Ar results from sanidine, plagioclase, and biotite in 16 tuff beds. Biotite symbols (diamonds) represent weighted mean plateau ages. MSWD—mean square of weighted deviates.
|
|
40Ar/39Ar Results
Sanidine
Fusions and incremental heatings were performed on single crystals and small multicrystal aliquots of sanidine from 18 of the dated samples (Figs. 6 and DR3). Single-crystal fusions of sanidine from six samples yielded precise results due to their large size (>250 µm), showed no distinct outliers, and gave weighted mean ages of 51.74 ± 0.09 Ma (Halfway Draw tuff), 47.70 ± 0.12 Ma (White Lignitic tuff), 48.76 ± 0.09 Ma (Church Buttes tuff), 48.62 ± 0.28 Ma (Leavitt Creek tuff), 48.15 ± 0.08 Ma (Henrys Fork tuff), and 48.11 ± 0.08 Ma (Tabernacle Butte tuff) (Fig. 6; Table 2). However, sanidine from several ash beds were too small (<180 µm in diameter) to achieve an adequate signal to blank ratio from single crystals, necessitating the use of multicrystal aliquots to achieve usefully high precision (i.e., <5% uncertainty). In most cases, multicrystal analyses yield Gaussian apparent age distributions with few outliers and give stratigraphically consistent weighted mean ages of 51.90 ± 0.09 Ma (Scheggs tuff), 50.61 ± 0.23 Ma (Boar tuff), 48.94 ± 0.29 Ma (Sand Butte tuff), 48.66 ± 0.28 Ma (Continental tuff), and 51.66 ± 0.09 Ma (K-spar tuff) (Fig. 6; Table 2). Note that new analyses of sanidine from the Firehole and Analcite tuff beds have resulted in a slight revision of the preferred ages for these beds (Table 2). Incremental heating experiments performed on multicrystal sanidine aliquots from several tuffs all yield internally concordant plateau ages consistent with fusion ages, suggesting that 40Ar* loss due to alteration is insignificant (Table 2; Fig. DR3).
Although concerns have been raised about the ability to distinguish xenocrysts when multiple sanidine crystals are analyzed (cf. Machlus et al., 2004; Smith et al., 2006), single-crystal fusion results suggest that contamination present in Green River Formation ash beds is composed of significantly older grains that can be readily distinguished and excluded even when using small multicrystal aliquots. To assess the distribution and magnitude of potential contamination of multicrystal aliquots, single-crystal fusions of sanidine from the Analcite, Scheggs, and Sixth tuffs were conducted. Although these measurements are relatively imprecise due to lower signal to blank ratios (i.e., >10% uncertainty), single fusions of Analcite tuff and Scheggs tuff sanidine produced no obvious outliers and yield weighted mean ages that are indistinguishable from those of multicrystal analyses (Figs. 6 and DR4). Sanidine from the Sixth tuff proved to be more problematic (cf. Smith et al., 2003). New single-crystal fusions (n = 37) of Sixth tuff sanidine yield a weighted mean of 48.68 ± 0.59 Ma that is consistent with the more precise age of 49.62 ± 0.10 Ma acquired via the incremental heating of individual biotite, which remains its preferred age (Smith et al., 2006). However, three single-crystal fusions gave apparent ages that are significantly overestimated (>70 Ma; Fig. DR4). Accordingly, when Sixth tuff sanidine were analyzed as 5 crystal aliquots (cf. Smith et al., 2003), only 11 of 30 fusions yielded stratigraphically reasonable apparent ages, whereas 19 gave distinctly older ages (Fig. DR4). We interpret these older outliers to reflect the admixture of a small proportion (<10%) of xenocrystic or detrital grains with a larger number of juvenile magmatic sanidine.
Sanidine from the Strawberry and Yellow tuff beds exhibited the largest amount of age scatter, which limits the precision of the age determinations for these beds (Fig. 7). More than 10% of single-crystal fusions of Strawberry tuff sanidine yielded distinctly older ages, and accordingly, >90% of 6- to 8-crystal analyses gave anomalously old apparent ages. Similarly, only 50% of 3-crystal fusions of sanidine from the Yellow tuff gave stratigraphically reasonable apparent ages, and only 20% of fusions of 20-crystal aliquots could be included in the weighted mean calculation. We infer that the age scatter observed for both samples can be attributed to the inclusion of 10%–20% inherited grains.
View larger version (26K):
[in this window]
[in a new window]
|
Figure 7. Cumulative probability diagrams showing sanidine and biotite ages obtained from the Yellow and Strawberry tuff beds. Feldspars from these beds exhibit marked contamination by older grains. Note that multiple single-crystal analyses isolate a young, presumably juvenile magmatic population from both ash beds when older xenocrysts are excluded from the age calculation. MSWD—mean square of weighted deviates.
|
|
Excluding the Strawberry and Yellow tuffs, only 11 of 296 (3.7%) of sanidine analyses were excluded from age calculations (Table 2; Fig. 6). Based on the apparent absence of fluvial reworking of most sampled tuffs (Table DR2; see footnote 1) and largely unimodal sanidine age distributions (Fig. 6; Table DR2), we interpret their weighted mean ages to represent the best estimate of their age of eruption and deposition.
Detrital Feldspar
Laser fusion measurements were performed on single detrital feldspar grains from three volcaniclastic sand beds and yielded apparent ages indicative of rapid transport and deposition of erupted materials with little admixture of detrital grains. Of 82 analyses of K-feldspar from the Antelope sand bed near the base of the Sand Butte bed of the Laney Member, 20 gave apparent ages that range from Paleocene to Proterozoic (Fig. 8). However, 62 analyses yield Eocene apparent ages that have a Gaussian distribution (Fig. 8), suggesting derivation from a common eruption or set of similarly timed eruptions. We take the weighted mean age of this younger component (48.70 ± 0.19 Ma) to indicate the age of this volcanism and the maximum age of sand deposition (cf. Deino and Potts, 1990). Similarly, 19 of 25 analyses of sanidine from the Sage tuff, a volcaniclastic sand bed in the Bulldog Hollow Member of the Fowkes Formation, gave internally consistent apparent ages with a weighted mean of 47.94 ± 0.17 Ma (Fig. 6). All 35 analyses of >0.5-mm-diameter sanidine from the Sage Creek Mountain pumice within a sand bed near the base of the Turtle Bluff Member of the Bridger Formation gave precise, consistent apparent ages that have a weighted mean of 47.17 ± 0.08 Ma (Fig. 6), suggesting their derivation from a single eruption or closely timed set of eruptions.
View larger version (17K):
[in this window]
[in a new window]
|
Figure 8. Cumulative probability diagram showing analyses of detrital orthoclase and sanidine from a sandstone bed near the base of the Sand Butte bed of the Laney Member. MSWD—mean square of weighted deviates.
|
|
Biotite
Several of the tuff beds sampled from the Uinta Basin lack sanidine, which necessitated the use of biotite. Multiple laser incremental heatings of hand-picked euhedral biotite crystals were conducted to assess the potential for alteration-derived discordance and inaccuracy. In addition, electron probe microanalysis was performed on biotite crystals to gauge the presence or absence of alteration phases. (Table DR4 and Figs. DR5 and DR6; see footnote 1) (cf. Smith et al., 2006). Age spectra produced from biotite from Green River Formation tuff beds are of two distinct types: (1) concordant and reproducible, and (2) discordant and indicative of alteration-related 40Ar loss and 39Ar recoil during irradiation. Discordant age spectra were observed from biotite from several tuffs (Fig. DR5), and correlate to the presence of intergrown alteration phases (Table DR4; Fig. DR6) and integrated age scatter toward both older and younger apparent ages (Fig. DR7). We consequently take the weighted mean of concordant biotite experiments as the best estimate for the eruptive age of the ash beds from which sanidine was unavailable (Fig. 6; Table 2), but caution that biotite populations yielding predominantly discordant spectra (i.e., the Curly, Wavy, Blind Canyon, and Fat tuffs) are less reliable than those yielding predominantly concordant plateaus, such as the Portly and Oily tuffs.
Analytical Results and Uncertainties
Because our initial emphasis is on delimiting the timing of lacustrine deposition in the Greater Green River Basin and Uinta–Piceance Creek Basin, 40Ar/39Ar ages are reported with 2 analytical uncertainties relative to the standard ages of Renne et al. (1998). Table 2 also reports intercalibration and fully propagated uncertainties for each unit (Karner and Renne, 1998; Renne et al., 1998). Intercalibration uncertainties should be considered when making comparisons to 40Ar/39Ar ages obtained using other standard minerals or to the geomagnetic polarity time scale (Fig. 9; Cande and Kent, 1992, 1995; Ogg and Smith, 2004). Fully propagated uncertainties reflect uncertainty in the 40K decay constant and K/Ar age of the GA-1550 primary biotite standard, and are required for comparisons with isotopic chronometers other than 40K decay, such as U-Pb (Karner and Renne, 1998; Min et al., 2000).
View larger version (42K):
[in this window]
[in a new window]
|
Figure 9. Integrated early and middle Eocene paleomagnetic chronostratigraphy for western North America. All 40Ar/39Ar ages are shown with 2 intercalibration uncertainties relative to the standard values of Renne et al. (1998).
|
|
|
ASH BED CORRELATIONS
|
|---|
Several possible correlations between ash beds are suggested by 40Ar/39Ar ages (Fig. 2B). Biotite phenocryst compositions provide an additional test of ash bed correlations (Desborough et al., 1973; Yen and Goodwin, 1976; Mauger, 1977). One potential correlation connects the Henrys Fork, Tabernacle Butte, and Wavy tuffs, all of which have overlapping 40Ar/39Ar ages (Table 2). The FeO/MgO and TiO2 compositions of biotite from the Henrys Fork and Wavy tuffs are similar (Fig. 10; Table DR4); however, biotite from the Tabernacle Butte tuff was not analyzed. At a broader spatial scale, these three ashes may correlate to the more proximal Blue Point Marker tuff in the southern Absaroka Volcanic Province, which was 40Ar/39Ar dated by Hiza (1999) as 48.10 ± 0.17 Ma (Fig. 5). Another potential correlation connects the Continental, Church Buttes, and Curly tuffs, which have overlapping 40Ar/39Ar ages and biotite with similar FeO/MgO and TiO2 compositions. However, the ages of the Continental and Curly tuffs also overlap with the age of the Leavitt Creek tuff, which is 100 m above the Church Buttes tuff but has not been analyzed for biotite composition (Fig. 10; Table 2).
View larger version (18K):
[in this window]
[in a new window]
|
Figure 10. FeO/MgO versus TiO2 plot of electron microprobe point analyses on biotite from select ash beds illustrating several likely correlations. Values are shown as weight percent. Biotite compositions for Henrys Fork and Church Buttes tuff beds are from Smith (2007). Lower TiO2 values of Desborough et al. (1973) and Mauger (1977) for biotite from the Wavy and Curly tuff beds may reflect the improved detection and interference correction capabilities of the Cameca SX51 electron microprobe and software used versus those utilized in the 1970s.
|
|
|
AGE MODEL FOR THE GREEN RIVER FORMATION
|
|---|
Calibration of North American Land-Mammal Ages
The 40Ar/39Ar ages presented here add to a radioisotopic data set (Wing et al., 1991; Smith et al., 2003, 2004, 2006) that provides temporal calibration for a large suite of existing lithostratigraphy, biostratigraphy, and magnetostratigraphy (Fig. 3). Overall, the ages determined for this study are entirely consistent with isochroneity of the North American land-mammal ages throughout the study area (Wood et al., 1941; Robinson et al., 2004). The direct temporal implications of these ages for the late Wasatchian, Bridgerian, and Uintan land-mammal ages are summarized in Figure 3 and detailed in Table 3.
Implications for Paleomagnetic Records and the Geomagnetic Polarity Time Scale
New 40Ar/39Ar ages allow for the calibration of early and middle Eocene paleomagnetic polarity records, which have been obtained at seven sites in basins containing the Green River Formation (Fig. 9; Jerskey, 1981; Flynn, 1986; Prothero and Swisher, 1992; McCarroll et al., 1996a; Stucky et al., 1996; Clyde et al., 1997, 2001). These studies have typically focused on alluvial strata where mammalian fossils are preserved, but in several cases have reported the polarity of lacustrine strata. Previous efforts to correlate magnetic polarity records to the geomagnetic polarity time scale have been hampered by a lack of common 40Ar/39Ar standard values (cf. Renne et al., 1998; Smith et al., 2003) and uncertain time-stratigraphic relationships between basins (cf. Wing et al., 2000; Smith et al., 2003; Machlus et al., 2004). However, when viewed in total (Fig. 9; Table 4), the current data set resolves many previous uncertainties and permits the recalibration, consistent with the standard ages of Renne et al. (1998), of the ages of chrons C24r through C20n. Our provisional calibration (Table 5) was calculated using seven 40Ar/39Ar ages for ash beds found within paleomagnetically characterized strata or their correlative equivalents (Fig. 9; Table 4). The new magnetic reversal ages (Table 5) are consistent with magnetostratigraphy at 19 locations, and with the presence of marine index taxa (P10, CP12b-13b) associated with C21n within the Ardath Shale; the shale underlies alluvial strata near San Diego, California, that contain Ui-1 faunas (Berggren et al., 1995; Walsh et al., 1996). The most significant resulting change involves a shift in the age of chron C22, which becomes 1 m.y. younger than indicated by either Cande and Kent (1992, 1995) or Ogg and Smith (2004). Another interesting feature is the presence of several brief polarity intervals that do not appear in seafloor magnetic anomaly records (Cande and Kent, 1992). If not the result of overprinting, such features may reflect short-term weakenings or reversals of the Earth's magnetic field similar to those observed in the Pliocene–Pleistocene record (Langereis et al., 1997; Singer et al., 2004).
Extrapolation of Age Model to Strata Not Directly Constrained by Tuff Beds
Ash beds have not been identified in the lowest portions of the Green River Formation below the Sheggs and Yellow tuffs (Fig. 2), and therefore mammalian biostratigraphy remains the most useful age constraint. For example, the alluvial DeBeque Formation, which is equivalent to the lower half of the Green River Formation in the Piceance Creek Basin, has yielded Graybullian (Wa-5) through Gardner-buttean (Br-0–Br-1a) faunas (Kihm, 1984; Froehlich and Froehlich, 2002). Strata equivalent to the lower part of the Green River Formation in the Greater Green River Basin have produced a similar faunal succession (Fig. 3; Holroyd and Smith, 2000; Zonneveld et al., 2000). In strata where fossil collections are absent, such as the Battle Spring Formation and Colton Formation (Spieker, 1946; Love, 1970), temporal assignments are typically based on bulk 
TOP
ABSTRACT
INTRODUCTION
