Provenance of the Paleogene Colton Formation (Uinta Basin) and Cretaceous–Paleogene provenance evolution in the Utah foreland: Evidence from U-Pb ages of detrital zircons, paleocurrent trends, and sandstone petrofacies Open Access

The Colton Formation is a succession of Upper Paleocene to Lower Eocene fluviodeltaic strata exposed in the Roan Cliffs of northeastern Utah (United States) along the southern flank of the Laramide Uinta Basin. Reaching a thickness of ∼1000 m in Desolation Canyon of the Green River, Colton fluviodeltaic strata interfinger northward in the basin subsurface with lacustrine strata of the Green River Formation. We use U-Pb ages for detrital zircons from Colton and related strata of the Uinta Basin in combination with sandstone petrofacies and paleocurrent trends to reach an appraisal of Colton provenance in the context of evolving Cretaceous–Paleogene depositional patterns in the Sevier retroarc foredeep and Laramide successor basins.

Assessment of Colton provenance provides a vehicle to highlight the utility of combining information from detrital zircons, petrofacies, and paleocurrents for interpretations of sediment provenance and dispersal. We also show the value of deconvolving detrital zircon populations into constituent subpopulations that can be treated separately, both on graphical plots and by using Kolmogorov-Smirnoff (K-S) statistics. The mixing of age subpopulations of detrital zircons from different provenances in varying proportions during sediment dispersal is an important aspect of provenance analysis that is not well elucidated by consideration of bulk detrital zircon populations without close attention to constituent subpopulations. Our analysis is based on U-Pb ages for 4765 detrital zircons in 57 sandstone samples, ∼8500 fluvial paleocurrent measurements, and 705 point counts of sandstone petrofacies from Cretaceous–Paleogene strata in the Utah–Arizona foreland of the Cordilleran backarc, and U-Pb ages for 905 detrital zircons in 54 sandstone samples from coeval strata in the California forearc.

In a preliminary study of Colton provenance (Davis et al., 2010), it was reported that half the detrital zircons in arkosic Colton sandstones have U-Pb ages younger than 285 Ma, implying that headwaters of the Colton dispersal system extended as far south as the Mojave segment of the Cordilleran magmatic arc, 750 km or more south of the Uinta Basin. These U-Pb data disproved the previous inference (Dickinson et al., 1986) that Colton arkosic sand was derived from Precambrian basement exposed in nearby Laramide uplifts. The fluvial mainstem for Colton sediment dispersal from the Cordilleran arc northward to the Uinta Basin is termed the California paleoriver (Fig. 1) (after Dickinson et al., 2011). We note that Zawiskie et al. (1982) postulated delivery of some significant fraction of fluviodeltaic Colton sediment to the Uinta Basin from as far south as central Arizona directly east of the Mojave region.

Table 1 lists the sandstone samples from Cretaceous–Paleogene strata of the Utah-Arizona foreland region for which U-Pb ages of detrital zircons are available. For U-Pb ages reported here for the first time, sample localities are provided in Supplemental File 1¹, U-Pb geochronological methods in Supplemental File 2², and full U-Pb analytical data in Supplemental File 3³, including both concordia diagrams and age-probability plots with age-bin histograms. Detrital modes for the same samples are provided in Table 2 as a guide to petrofacies.

U-Pb ages from most samples (86%) were determined by laser ablation–inductively coupled plasma–mass spectrometry in the Arizona LaserChron Center at the University of Arizona (Gehrels et al., 2008; Dickinson and Gehrels, 2009). U-Pb ages for eight samples were determined by secondary ion mass spectrometry (SIMS) technology at the Australian National University (Larsen et al., 2010). The age spectrum for one of the SIMS samples is incompatible with other data, perhaps because of stratigraphic miscorrelation, and was not used for statistical analyses of age data (Table 1). U-Pb ages from most samples (91%) were determined for randomly selected detrital zircon grains, but the dated grains from five of the samples were selected preferentially as the most limpid and euhedral grains in a deliberate search for the youngest grains present (Mathers, 2009). Those samples were used for some but not all statistical analyses because the full age spectra of their detrital zircon populations are inconsistent with data for other samples from the same stratigraphic units.

We use three complementary criteria for comparison of detrital zircon age populations or subpopulations.

(1) Age distribution curves in the form of age probability plots (Ludwig, 2003) for which each grain age is cast as a normal distribution including its standard deviation of age error, and all the normal distributions for individual grain ages are then summed into curves normalized to subtend equal areas below the curves. Peaks on age distribution curves are statistically rigorous representations of analytical data, but measured age peaks do not necessarily reflect faithfully the actual distribution of grain ages within a zircon concentrate because the randomly selected grains dated are but a sampling of the total grain populations. Andersen (2005) showed that age peaks on age distribution curves are dependent in detail on the random selection of grains dated, as well as on the distribution of grain ages in the population sampled.

(2) Probability (P) values calculated from Kolmogorov-Smirnoff (K-S) statistics (Press et al., 1986). To supplement visual inspection of age distribution curves, and to shed light on the influence of random grain selection for the configurations of the curves, we use K-S statistics to test the null hypothesis that two detrital zircon age populations or subpopulations might have been selected at random from the same parent population. Where P > 0.05 from K-S analysis, with the analytical uncertainties of each grain age taken into account, one cannot infer with 95% confidence (0.95 being the inverse of 0.05) that 2 detrital zircon age populations were not selected at random from the same parent population (with P = 1.0 indicating statistical identity). We conclude that no robust provenance distinctions can be inferred for two detrital zircon age spectra yielding P > 0.05 from K-S analysis. Even where P < 0.05 but age peaks on age distribution curves are the same, 2 contrasting age spectra may reflect derivation of detrital zircon subpopulations in varying proportions from the same source rocks within a common provenance.

(3) Tabulated age subpopulations (Table 3) of detrital zircons based on best estimates of grain age ignoring analytical uncertainties. This approach is not statistically rigorous but allows numerical manipulation to calculate grain age indices indicative of key provenance contrasts. Age boundaries between the age subpopulations are based empirically upon nulls in the patterns of grain ages for the samples of Table 1, but the peaks for the subpopulations derived from age distribution curves are better guides to the central age span of each subpopulation (Table 3). The arc-derived subpopulation (I) is delimited by the maximum age (ca. 285 Ma) of the oldest igneous assemblages of the Cordilleran magmatic arc (Dickinson and Gehrels, 2009). We refer to older detrital zircons as pre-arc grains.

Detrital zircon samples (Table 1) derive from 20 different stratigraphic units of the Utah foreland (Fig. 2). Colton samples are among the richest (∼50%) in arc-derived detrital zircons.

The fluviodeltaic Colton Formation (Spieker, 1946) of the Uinta Basin underlies the lacustrine Green River Formation (Fig. 3) into which Colton strata intertongue laterally and grade upward (Fouch, 1976), and overlies the Flagstaff Limestone (Spieker, 1946) of lacustrine origin, now considered a basal member of the Green River Formation (Davis et al., 2009b). Colton strata include both alluvial and delta-plain deposits (Morris et al., 1991), including tan to red channel-form sandstone bodies and intervening intervals of red to green overbank mudstone; the proportions of channel and overbank deposits are areally variable. Arkosic sandstone is dominant in the core of the Colton sediment wedge at the Green River. Lithologically similar strata beneath the Green River Formation in the Piceance Basin (Fig. 1) are termed the DeBeque Formation (Smith et al., 2008).

The Colton Formation is exposed continuously for ∼200 km along the face of the Roan Cliffs, and extends downdip into the subsurface of the Uinta Basin where it grades northward into the Green River Formation (Fig. 3). Modeling Colton sediment volume as a half-cone with its apex at Desolation Canyon of the Green River, where the thickness is ∼1000 m (Cashion, 1967), yields ∼3000 km³ of preserved Colton sediment. The original sediment volume was greater because no regional thinning of Colton strata southward toward surface outcrops has been detected (Fouch et al., 1976, 1992; Franczyk et al., 1989). Paleocurrents indicate derivation of Colton sediment consistently from southern azimuths, whereas paleocurrent trends for underlying units reflect derivation from western azimuths (Fig. 4).

For 75–85 km along the southeastern flank of the Uinta Basin, basal Colton Formation includes 10–50 m of pebble conglomerate and conglomeratic sandstone of the Paleocene Dark Canyon sequence (Figs. 3 and 4), a braidplain succession that unconformably overlies the Cretaceous (Campanian) Tuscher Formation, which is commonly bleached below the contact (Willis, 1986; Franczyk and Pitman, 1987; Franczyk et al., 1990). Sandstone-rich intervals at the top of the Dark Canyon sequence locally grade upward into the Colton Formation (Mathers, 2009). In other places, sharp transitions between conglomerate and sandstone reflect condensation of section within an evolving fluvial succession. Dark Canyon clasts are chert and quartzite pebbles probably derived from Mesozoic strata eroded off the growing Uncompahgre uplift to the southeast (Fig. 1), from which Dark Canyon paleocurrents (Fig. 4) suggest derivation of sediment.

Along the southwestern flank of the Uinta Basin, lithologically analogous pebbly beds (Lawton, 1983, 1986b), 6–18 m thick and composed dominantly of chert and quartzite pebbles dispersed in pebbly sandstone, are present unconformably beneath Paleocene strata of the North Horn Formation, where they concordantly overlie the Cretaceous (Campanian) Farrer Formation on the plunging nose of the San Rafael Swell (Fig. 4). The pebbly beds are interpreted as a lateral vestige of the Tuscher Formation (Lawton, 1983, 1986b), which was otherwise removed by erosion across the growing San Rafael Swell beneath the unconformity with the North Horn Formation (Fig. 4). We infer that the clasts in the pebbly beds were derived from Mesozoic strata of the Laramide San Rafael Swell as it began to grow in the Sevier foreland before it had developed enough structural relief to interrupt sedimentation.

Interfingering of the Colton Formation with the overlying Green River Formation led to the progradation of deltaic bodies as stratigraphic tongues projecting into lacustrine successions. The Sunnyside delta complex (Remy, 1992; Schomacker et al., 2010) is a lateral equivalent of the uppermost part of the Colton Formation as exposed farther east along the Green River (Fig. 4). The Horse Bench Sandstone Bed of the Green River Formation (Fig. 4) is the stratigraphically highest record of delta-related sediment of arkosic petrofacies (Table 2) in the Uinta Basin, and maintains a consistent thickness of ∼6 m for long distances along strike west of the Green River. East of the Green River, the Horse Bench Sandstone Bed is <6 m thick, and is mapped as the base of the Evacuation Creek Member overlying the Parachute Creek Member (Cashion, 1967).

East of the Green River, the Colton Formation thins as it onlaps the flank of the Laramide Douglas Creek arch (Fig. 4). Only thin sandstone intervals mapped either below or within the basal Green River Formation are present on the crest of the arch as lateral equivalents or younger analogues of the Colton Formation to the west in the Uinta Basin and the DeBeque Formation to the east in the Piceance Basin (Fig. 3). Paleocurrents in the basal Paleogene sandstones on the Douglas Creek arch diverge to the northwest and northeast off the crest of the arch (Fig. 4), suggesting that Colton and DeBeque depositional systems were separate. DeBeque paleocurrents toward the north-northwest in the Piceance Basin (Lorenz and Nadon, 2002) imply that Colton sediment was not transported across the Douglas Creek arch, although post-Colton Lake Uinta eventually overtopped the arch.

In the San Pitch Mountains (Gunnison Plateau) and Cedar Hills ∼75 km southwest of the Uinta Basin (Fig. 3), Paleogene fluviatile sandstone and associated floodplain mudstone and siltstone intervening between lacustrine strata of the Flagstaff Limestone Member below and the main body of the Green River Formation above reach thicknesses of 165–245 m, and have been mapped as the Colton Formation (Marcantiel and Weiss, 1968; Witkind et al., 1987; Witkind and Weiss, 1991). The exposures are not contiguous, however, with the Colton Formation of the Uinta Basin, and the quartzose sandstones of the succession are unlike the Colton arkosic petrofacies (Table 2). In the absence of an alternate name for the so-called Colton Formation of the San Pitch Mountains, we refer to the strata as “Colton” Formation because they have been mapped to date only as Colton Formation, but in our view, merit a separate stratigraphic name.

For provenance analysis, 52 of the 57 detrital zircon samples of Table 1 are grouped into 13 stratigraphic and areal subsets (A–M, Table 4). Samples DR (Dome Rock), COL5 (North Horn), GRF (Flagstaff), COL9 (“Colton”), and 12JL5 (Claron) are treated individually. From K-S analysis, P > 0.05 for comparisons of samples within each subset except for subsets A and L, for which we infer that composite detrital zircon populations for multiple samples are nevertheless more reliable indicators of net provenance than the populations of individual samples.

Sediment was delivered to the Cretaceous Sevier foredeep of the southern Cordilleran foreland basin both by transverse paleoflow eastward off the Sevier thrust belt and by longitudinal paleoflow northward from the Mogollon highlands transecting central Arizona (Lawton et al., 2003; Dickinson and Gehrels, 2008, 2010a, 2010b; Lawton and Bradford, 2011). The Sevier and Mogollon provenances yielded contrasting pre-arc detrital zircon populations. Detritus from the Mogollon provenance (Fig. 5D) was dominated by subpopulations IV (anorogenic granite) and V (Yavapai-Mazatzal) of Table 3, whereas detritus from the Sevier provenance (Fig. 5ABC) was dominated by subpopulations II (Paleozoic–Neoproterozoic) and III (Grenville) recycled from Paleozoic and Mesozoic strata of the thrust belt (Dickinson and Gehrels, 2008; Larsen et al., 2010).

Subordinate proportions of subpopulations IV and V are present in Sevier detritus because those subpopulations are present in stratigraphic units of the thrust belt and were recycled together with subpopulations II and III (Dickinson and Gehrels, 2008; Lawton et al., 2010). The reduced proportion of subpopulations IV and V for the southern Sevier provenance (Fig 5C), as opposed to the northern Sevier provenance (Fig. 5AB), is interpreted to reflect derivation from a different mix of strata within the thrust belt. The age range of recycled detrital zircons varies longitudinally in the Cordilleran foreland basin because the detrital zircon populations of strata incorporated into the thrust belt vary along strike (Leier and Gehrels, 2011).

The paucity of arc-derived detrital zircons in sand spread eastward into the foredeep from sources along the Sevier thrust belt implies that arc-derived detritus reaching the Utah foreland did not travel transversely across the Sevier orogen from the Sierra Nevada segment of the Cordilleran magmatic arc directly to the west, but instead moved longitudinally along the foredeep east of the thrust front from segments of the arc farther south beyond the southern limit of Sevier thrusting (Fig. 1). Intramontane Cretaceous–Paleogene deposits within the Sevier orogen between the Sierra Nevada and the Sevier thrust belt contain few arc-derived detrital zircon grains (Druschke et al., 2011). Minor detrital zircon grains of Jurassic and Cretaceous age may have reached the foredeep from backarc igneous centers (du Bray, 2007) located east of the main arc trend along the Sierra Nevada.

For analysis of Cretaceous–Paleogene depositional patterns in the backarc foreland of Utah, we employ two grain age indices. The arc index (100 × subpopulation I of Table 3/total detrital zircons) indicates the proportion of arc-derived grains in detrital zircon populations, and ranges from 0 to ∼75 (an arc index of 100 would indicate arc-derived grains only). The Mogollon index (subpopulations IV + V/total pre-arc grains) indicates the proportion of pre-arc grains derived from the Paleoproterozoic Yavapai-Mazatzal belt of southwest Laurentia as intruded by anorogenic granites of Mesoproterozoic age, and ranges from 20 to ∼95 (a Mogollon index of 100 would indicate pre-arc grains exclusively of subpopulations IV and V).

A plot of arc index against Mogollon index indicates the relative contributions of detrital zircon grains from the two provenances in relation to contributions from the magmatic arc (Fig. 6). A Mogollon index of 60 delimits the field of Sevier-derived detritus with arc indices <10. For samples with a Mogollon index <60, the plot suggests that arc-derived grains were added to pre-arc detrital zircon populations derived largely if not exclusively from the Sevier provenance. The ancestry of individual grains of subpopulations IV and V is equivocal, however, because grains derived directly from the Mogollon highlands cannot be distinguished in derivative sandstones from grains derived initially from southwest Laurentia but recycled from strata uplifted along the Sevier thrust belt.

For a Mogollon index >60, arc-derived grains were by inference added to pre-arc detrital zircon populations containing significant proportions of detritus from the Mogollon provenance. Mogollon indices in the range of 60–95 imply admixtures of detritus from Sevier and Mogollon provenances in varying proportions, with the maximum Mogollon index of ∼95 observed only for subset M (Table 4), the Mogollon provenance signature of Figure 5D, and sample DR (Dome Rock Upper Cretaceous of Table 1) collected south of the limit of Sevier thrusting (Fig. 1). The Mogollon index of subset B (Colton) is intermediate between Mogollon indices for Sevier and Mogollon provenances (Fig. 6), suggesting that detritus from both provenances reached the California paleoriver carrying Colton sediment (Fig. 1).

For the Utah foreland, arc indices >30 are observed only for strata of late Campanian and younger age and Mogollon indices >60 are observed only for Paleogene and the youngest preserved late Campanian (Tuscher) strata (Fig. 2). These relations suggest coupled enhancement over time of both arc and Mogollon detritus transported longitudinally from south to north within the foreland region.

K-S comparisons of total detrital zircon populations in the sample subsets of Table 4 and selected individual samples (Table 1) having an arc index >10 show that P < 0.05 for 93% of sample pairs (Table 5), largely because variable proportions of arc-derived and pre-arc grains reduce P values. P is high (>0.25) only for comparisons of subset B (Paleogene Colton) with sample DR (Upper Cretaceous from the Mojave region), a relation noted in Davis et al. (2010), and of subset F (Campanian Kaiparowits) with sample GRF (Paleogene Flagstaff underlying Colton). Separate K-S analyses of arc-derived and pre-arc subpopulations help to clarify provenance relations.

For arc-derived grains considered separately from pre-arc grains in the samples of Table 5, P > 0.05 for 78% of sample pairs (Table 6), suggesting that age spectra of detrital zircons in arc detritus delivered to the backarc Utah foreland were largely comparable during Late Cretaceous and early Paleogene time. Note, however, that P < 0.05 for comparisons of Paleogene Colton of the Uinta Basin (subset B) with Paleogene DeBeque of the Piceance Basin (subset C), and with most Cretaceous subsets (D–F, K, L).

For pre-arc grains in most samples of Tables 5 and 6, P > 0.05 for 34% of sample pairs (Table 7), more than for total grain populations (10%) but fewer than for arc-derived grains (73%) in the same samples. Arc-derived grains having similar age spectra were evidently added to pre-arc grains of varied Sevier or Mogollon or mixed provenance. Note that P > 0.05 for comparisons of Colton (subset B) with the underlying Tuscher Formation (subset D) and for the Farrer Formation (subset E), which underlies the Tuscher Formation, with the laterally equivalent Kaiparowits Formation (subset F) of the Table Cliff basin. Conversely, P < 0.05 for either Colton or Tuscher grains in comparison with either Farrer or Kaiparowits grains. These relations suggest progressive provenance evolution in the Utah foreland over the interval spanning Kaiparowits–Farrer and Tuscher–Colton Formations deposition (Fig. 2).

K-S comparisons for pre-arc grains in samples with contrasting Mogollon indices (>60 and <60) consistently yield P ≤0.05 and are not tabulated. For pre-arc grains in sample subsets and individual samples of varied Cretaceous–Paleogene ages but with Mogollon index <60, P > 0.05 for only 38% of sample pairs (Table 8), suggesting mixing in varying proportions of detritus from Sevier and Mogollon provenances in varying patterns over time within the Utah foreland.

Detrital zircon populations from Paleogene strata of the Uinta Basin and nearby areas (Figs. 3 and 4) fall into three groups reflecting different provenance relations (Fig. 7): (1) samples from the North Horn and “Colton” Formations that contain ≤1% arc-derived grains, (2) samples from the Colton Formation, including the Dark Canyon sequence, that contain 44%–58% arc-derived grains, and (3) samples from the DeBeque Formation and Sunnyside delta complex, including the Horse Bench Sandstone Bed, that contain an intermediate level of 12%–23% arc-derived grains.

The Maastrichtian–Paleocene North Horn Formation (COL5 of Table 1) exposed along the southwest flank of the Uinta Basin and the Paleocene–Eocene “Colton” Formation (COL9 of Table 1) in the San Pitch Mountains southwest of the Uinta Basin (Figs. 2 and 3) contain essentially no arc-derived detrital zircons (Fig. 8AB), and P = 0.81 from K-S comparison of their full age spectra. The age spectra of pre-arc grains in both samples are similar to those for sandstones derived mainly from the northern Sevier thrust belt (Figs. 5D and 8C), with P = 0.14 (“Colton”) and P = 0.52 (North Horn) from K-S comparisons of pre-arc grains (Table 8). Facies relations in the San Pitch Mountains suggest that “Colton” sand was derived exclusively from nearby frontal Sevier thrust sheets (Marcantiel and Weiss, 1968), and North Horn paleocurrents (Fig. 4) are compatible with derivation from the Sevier thrust belt to the west.

The Paleogene Colton Formation and Cretaceous strata of the Utah foreland contain similar arc-derived detrital zircon subpopulations (Fig. 9), although proportions of arc-derived grains are variable. All three arc-derived subpopulations (Ia–Ic of Table 3) are present. Samples from the conglomeratic Dark Canyon sequence (basal Colton Formation) were composited with other Colton samples because our single sample from the Dark Canyon sequence (COL8 of Table 1) yields P = 0.39–1.00 for all grains, P = 0.85–1.00 for arc-derived grains, and P = 0.51–0.99 for pre-arc grains when compared by K-S analysis to our other four Colton samples, and the single Mathers (2009) Colton sample from above the Dark Canyon sequence (DC3 of Table 1) yields P = 0.31–0.99 from comparison with her three Dark Canyon samples.

Upper Campanian strata (Fig. 9BCD) in the Utah foreland contain in net as many arc-derived grains (∼50%) as the Colton Formation (Fig. 9A), and distinctly more than older Upper Cretaceous strata (Fig. 9E). Paleocurrents and facies patterns in the pre-upper Campanian strata reflect components of longitudinal paleoflow from the southwest in the Utah foreland (am Ende, 1991; Lawton et al., 2003; Garrison and van den Burgh, 2004; Janok et al., 2010), even though the net proportion of arc-derived grains is much lower than in upper Campanian units. These relations suggest that some arc-derived detritus began to reach the backarc region of Utah with the initiation of the Cordilleran foreland basin in mid-Cretaceous time, but not in abundance until Campanian time, after which delivery of abundant arc detritus to the Utah foreland continued into Paleogene time. Fluvial transport of arc-derived grains into the Uinta Basin from the south by longitudinal rather than transverse paleoflow with respect to the Sevier thrust belt to the west is inferred because units containing abundant arc-derived zircons (>25%) display northeastward to northwestward paleocurrents, whereas units lacking many arc-derived zircons (<10%) display eastward to southeastward paleocurrents reflective of sediment transport off the Sevier thrust belt (Fig. 10).

No strata of Maastrichtian and Early Paleocene age in the Uinta Basin (Fig. 2) contain abundant arc-derived zircons. The Upper Paleocene to Lower Eocene Colton Formation (Fig. 2) either unconformably overlies subjacent Campanian strata or is separated from the Campanian strata by the intervening Sevier-derived sediment wedge of the Maastrichtian to Lower Paleocene North Horn Formation (Fig. 4). There is no internal evidence within the Uinta Basin for the persistence of longitudinal foreland transport of arc detritus between late Campanian and Late Paleocene time during peak Laramide deformation.

The Colton arkosic petrofacies represented the culmination of a persistent trend of compositional evolution for sandstones of the Uinta Basin (Fig. 11). Quartzolithic sandstones derived from the Sevier thrust belt were progressively supplanted by increasingly feldspathic sandstones as the proportion of arc detritus reaching the basin increased during late Campanian (Farrer and Tuscher) and Paleogene (Colton and Green River) time. North Horn sedimentation reflected a transient return to the Sevier-derived quartzolithic petrofacies in Maastrichtian to Early Paleocene time. Cretaceous Indianola and Paleogene “Colton” sandstones deposited near the thrust front contain petrofacies intermediate between Sevier-derived quartzolithic and Colton arkosic petrofacies.

The age spectra of pre-arc detrital zircon grains in Colton and older sandstones of the Utah foreland reinforce the interpretation that longitudinal sediment transport along the trend of the foreland basin increased in importance over Cretaceous–Paleogene time (Fig. 12). The Mogollon index (Fig. 7) is <60 for the lower three curves but >60 for the upper two curves, which also reflect reduced contributions of detrital zircon subpopulations II and III recycled from the Sevier thrust belt. For Tuscher and Colton subsets (Figs. 12A, 12B) of most arkosic composition (Fig. 11) and highest Mogollon index (Fig. 7), feldspar may well have been derived from Precambrian basement of the Mogollon provenance as well as from plutons or volcanic assemblages of the Cordilleran arc in the Mojave region. The close similarity of Kaiparowits (Fig. 12D) and Farrer (Fig. 12C) age spectra for pre-arc grains, and their contrast with mutually similar Tuscher (Fig. 12B) and Colton (Fig. 12A) age spectra, parallel the results obtained from K-S comparisons of those sample subsets (Table 7).

Detrital zircon populations in the conglomeratic Dark Canyon sequence at the base of the Colton Formation and within the overlying main body of the Colton Formation are similar (Table 4), despite contrasts in comparative lithology. The gravel fraction of the Dark Canyon sequence is a mature pebble lag apparently reworked from underlying upper Campanian Tuscher Formation beneath the basal Colton unconformity (Fig. 4). The sand fraction of the Dark Canyon sequence is less feldspathic and more lithic than the Colton arkosic petrofacies (Fig. 11), with a significant proportion of chert and quartzite sand grains (Table 2) presumably related to the pebble fraction of the Dark Canyon sequence.

Reworking of Colton sand from the Tuscher Formation is not favored for several reasons: (1) the arc-derived detrital zircon subpopulation (I of Table 3) is more abundant by a factor of nearly two in Colton sand as compared to Tuscher sand (44%–58% vs. 28%–31%), and zircons could not be sorted so strongly by age during recycling; (2) Colton arkosic sandstone is consistently more feldspathic than Tuscher sandstone (Fig. 11), whereas recycling is expected to enhance, rather than reduce, the ratio of quartz to feldspar; and (3) the Colton Formation is five times as thick as the Tuscher Formation (Fig. 4), which in any case was masked from erosion within the Uinta Basin once the Dark Canyon sequence was deposited above it.

To explain the close similarity of the detrital zircon populations in the basal Dark Canyon sequence and the remainder of the Colton Formation, we suggest a sedimentological rationale by which locally derived chert-rich and pebbly detritus was mixed on a conglomeratic braidplain with arkosic sand of distal origin during the earliest phase of Colton sedimentation. Intimate intercalation of lenses of cross-bedded sandstone and massive pebble conglomerate within the Dark Canyon sequence are compatible with that viewpoint. The similarity of Colton and Tuscher pre-arc subpopulations (Figs. 12A, 12B with P = 0.58 from Table 7) indicates that contamination of Colton sand with reworked Tuscher sand in the Dark Canyon sequence could not be readily detected from ages of detrital zircons. The greater abundance of arc-derived detrital zircons in the Colton sand would tend to overprint their lesser proportion in reworked Tuscher sand, which contains a higher ratio of subpopulation Ia (Cretaceous) to subpopulation Ib (Jurassic) grains within an arc-derived subpopulation of the same overall age span (Figs. 9A, 9B).

The age spectra of arc-derived (Fig. 13) and pre-arc (Fig. 14) detrital zircon populations in stratigraphic units overlying and laterally equivalent to the Colton Formation delineate the subregional extent of the Colton arkosic petrofacies. Two sandstones from the Green River Formation (Fig. 3) were collected from interbedded lacustrine and deltaic strata (Sunnyside delta) capped by the Horse Bench Sandstone Bed (Fig. 4). They are composited together (sample subset A of Table 4) as representative of arkosic sand (Table 2; Fig. 11) that was dispersed into the lacustrine Green River Formation of the Uinta Basin as deltaic distributary channels and mouth bars (Schomacker et al., 2010) spread laterally into Lake Uinta from Colton-like sources. P < 0.05 for K-S comparisons of total detrital zircon populations in subset A and the other subsets of Figure 13 from the Uinta Basin (Table 5), but the contrast stems in large part from different proportions of arc-derived grains, for which P = 0.18–0.49 for the same samples (Table 6). Arc-derived subpopulations are less abundant (14%–27%) in Sunnyside delta–Horse Bench (Fig. 13A) and DeBeque (Fig. 13B) samples than in the other sample subsets (38%–75%) of Figure 13. P = 0.01 (Table 7) for pre-arc subpopulations (Fig. 14AB) in sample subsets A (Green River) and B (Colton), suggesting that sandstones in the lower Green River Formation of the Uinta Basin represent mixtures of Colton-like arkosic detritus (Table 2 and Fig. 11) rich in arc-derived detrital zircons with pre-arc grains delivered to Lake Uinta from more proximal sources in surrounding Laramide uplifts.

For arc-derived grains in sandstones of the Green River Formation in the Piceance Basin (Davis et al., 2009b), P = 0.00 from K-S comparison with Colton sands of the Uinta Basin but P = 0.90 from K-S comparison with DeBeque sands of the Piceance Basin (subset C of Table 4), confirming that sediment transported into the Uinta Basin by the California paleoriver was not carried across the Douglas Creek arch into the Piceance Basin (Fig. 1). Note that ca. 50 Ma detrital zircons delivered during the interval 49.5–47.0 Ma to the Green River Basin north of the Uinta uplift (Figs. 1 and 3) by the Idaho paleoriver (Chetel et al., 2011) from sources in the Challis volcanics were not detected in the arkosic Horse Bench Sandstone Bed (ca. 47.5 Ma) at Gate Canyon (Table 2; Fig. 4), or in sandstones from the Green River Formation in the Piceance Basin (Davis et al., 2009b). Volcaniclastic detritus derived from the north (Surdam and Stanley, 1980) evidently did not reach basins south of the Uinta uplift until after ca. 47.5 Ma.

P ≤0.01 for K-S comparisons of total, arc-derived, and pre-arc detrital zircons in Colton and DeBeque sample subsets (Tables 5–7), in keeping with fluvial paleoflow to the northwest out of Colorado for the DeBeque Formation (Lorenz and Nadon, 2002), rather than from Utah to the west. The prominence of 60–80 Ma age subpeaks for DeBeque grains (Fig. 13B) that are absent from Colton (Fig. 13C) grains probably reflect derivation of Cretaceous detrital zircons in DeBeque sand from igneous rocks of the Colorado Mineral Belt (Fig. 1), where ages of 75–65 Ma for intrusions are characteristic (Chapin, 2012). The Yavapai-Mazatzal age peak for DeBeque (subpopulation V of Fig. 14B) was probably derived from Precambrian basement in Laramide uplifts of Colorado rather than having Mogollon provenance, and the DeBeque Yavapai-Mazatzal age peak is slightly older than the Yavapai-Mazatzal age peaks for Colton and related strata in Utah. The DeBeque pre-arc age subpeak at 1885 Ma is not present on any of the age spectra for strata deposited in Utah (Fig. 14), and may be a signal of Colorado rather than Mogollon provenance within the Yavapai-Mazatzal belt.

Subpeaks in the range of 70–85 Ma for arc-derived subpopulation 1a (Table 3) in Farrer–Tuscher (Fig. 13E) and Flagstaff (Fig, 13D) samples are absent from the Colton samples (Fig. 13C), but are also prominent for the Kaiparowits Formation of the Table Cliff basin (Fig. 9D). The similarity of Kaiparowits and Farrer–Tuscher detrital zircon populations can be attributed to an upstream-downstream relationship between Kaiparowits and Farrer depositional systems (Lawton and Bradford, 2011). Field evidence for possible reworking of Kaiparowits sand into the Flagstaff depositional system is hidden beneath the Marysvale volcanic field (Fig. 1), but feldspathic sandstones (Stanley and Collinson, 1979) in the Flagstaff Limestone of the Wasatch Plateau north of the volcanic field may reflect a sedimentological connection.

Based on his interpretation that the California paleoriver reversed course through the Grand Canyon region (Fig. 1) from northeastward Cretaceous paleoflow to southwestward Paleogene paleoflow, Wernicke (2011) suggested that Colton detritus was recycled from Kaiparowits-equivalent strata exposed in Laramide uplands of southern Utah; his hypothesis cannot be tested directly because the only Kaiparowits-equivalent strata known south of the Uinta Basin in the Utah foreland, other than exposures of the Kaiparowits Formation in the Table Cliff basin (Fig. 15), form a thin succession (∼30 m) of fine-grained strata at the erosional top of the Cretaceous section preserved in the Henry Mountains basin (Fig. 1). Detrital zircon populations, sandstone petrofacies, paleocurrent trends over time, relative sediment volumes, and the timing of Kaiparowits stripping off the Kaibab uplift do not favor recycling of Colton arkosic sediment from Cretaceous strata.

Colton and Kaiparowits detrital zircon populations yield P = 0.00 for total, arc-derived, and pre-arc grains (Tables 5–7). This result is atypical for K-S analysis of recycled detrital zircons and their source strata. For example, P = 0.59 for pre-arc subpopulations in Lower and Middle Jurassic eolianites of the western Colorado Plateau and in Upper Jurassic sandstones of the Morrison Formation derived from recycling of the older eolianites from the Sevier thrust belt (Dickinson and Gehrels, 2008), and P = 0.69 for pre-arc subpopulations in Middle to Upper Jurassic eolianites of the eastern Colorado Plateau and in Lower Cretaceous sandstone of the Cintura Formation in southeastern Arizona derived from recycling of older eolianites uplifted along the Mogollon highlands rift shoulder of the Bisbee basin (Dickinson et al., 2009).

Arc-derived grains are more abundant in Kaiparowits samples (50%–90%) than in Colton samples (44%–58%), with a net Kaiparowits content of 74% and a net Colton content of 52% (Fig. 9). Mogollon indices are 66 for Colton but only 41 for Kaiparowits (Figs. 6 and 12), reflecting derivation of Kaiparowits pre-arc subpopulations dominantly from the Sevier thrust belt to the west with minimal input from the Mogollon highlands tapped by the California paleoriver farther south (Fig. 1). Paleocurrent trends in the Kaiparowits Formation are eastward (Fig. 16), reflecting derivation of detritus from near the southern terminus of the Sevier thrust belt rather than from the central Mojave headwaters region for the California paleoriver (Fig. 1).

For arc-derived grains, the principal subpeaks for subpopulation Ia (Cretaceous) are 75–80 Ma for Kaiparowits grains but ca. 95 Ma for Colton grains, and for subpopulation Ib (Jurassic) are 152 Ma for Kaiparowits grains and 168 Ma for Colton grains (Figs. 9A, 9D). The content of subpopulation Ic (Permian–Jurassic) is proportionally four times as great in Colton sand (16% of subpopulation I) as in Kaiparowits sand (4% of subpopulation I). We infer that arc-derived Cretaceous and Jurassic detrital zircons in Kaiparowits and Colton sands had their sources in different segments of the Cordilleran magmatic arc exposing a different mix of Mesozoic igneous rocks, and that the higher proportion of the Permian–Triassic subpopulation Ic in Colton sand reflects derivation from farther south and east, where Permian and Triassic components of the Cordilleran igneous assemblage are more prominent.

Kaiparowits sandstones are uniformly more lithic than Colton sandstones (Fig. 17). The largely volcaniclastic Kaiparowits petrofacies contrasts more strongly with the Colton arkosic petrofacies than does any other Cretaceous–Paleogene sandstone suite of the Utah foreland other than the quartzolithic petrofacies derived from recycling of sedimentary detritus from the Sevier thrust belt (Figs. 11 and 17). Destruction of 50%–75% of lithic fragments in sand by preferential abrasion during recycling, while maintaining the quartz to feldspar ratio unchanged, seems unlikely and an unnecessary postulate given the comparative Kaiparowits-Colton detrital zircon data.

The Kaiparowits Formation in the Table Cliff syncline is approximately as thick (∼1000 m) as the Colton Formation at Desolation Canyon of the Green River (Fig. 4), but is less extensive laterally (Fig. 1), with preserved sediment volumes ∼3000 km³ for the Colton versus ∼500 km³ for the Kaiparowits. All preserved Kaiparowits Formation remnants were buried beneath latest Cretaceous and earliest Paleogene strata at the time of Late Paleocene–Early Eocene Colton sedimentation (Fig. 15). The hypothesis that Kaiparowits sand was recycled into the Colton Formation thus requires the supposition that Kaiparowits-equivalent strata were once present over large areas to the east of the Table Cliff syncline, where they were not capped by younger strata but were later removed by Paleogene erosion.

Kaiparowits strata in the Table Cliff syncline constitute a distal fluvial facies of meander belt and anastomosed alluvial deposits (Eaton et al., 1987; Goldstrand, 1992; Lawton et al., 2003; Roberts, 2007; Lawton and Bradford, 2011). Only the lowermost Kaiparowits Formation preserves any record of tidal influence on sedimentation from the Cretaceous interior seaway, and the bulk of the unit may have formed as a terminal fluvial accumulation deposited rapidly within an incipient Laramide down-bowing on the site of the Table Cliff syncline. Kaiparowits-equivalent strata in the Henry Mountains basin to the east are gray mudstone with minor thin sandstone interbeds (called “beds on Tarantula Mesa” by Peterson and Ryder, 1975) that overlie the Tarantula Mesa Sandstone (Eaton, 1990; Roberts et al., 2005; Jinnah and Roberts, 2011). The Tarantula Mesa Sandstone is a correlative of the capping sandstone member of the Wahweap Formation (Figs. 2 and 17) that underlies the Kaiparowits Formation in the Table Cliff syncline. No sandstone-rich Kaiparowits equivalents are known from any locale east of the Table Cliff syncline and south of the Uinta Basin, where the Farrer Formation contains similar detrital zircon subpopulations (P = 0.55 for arc-derived grains and P = 0.36 for pre-arc grains; Tables 6–8).

The distal character of Kaiparowits fluvial facies in the Table Cliff syncline, paleocurrents from the southwest, and detrital zircons reflecting derivation of Kaiparowits sand from both the Sevier thrust belt and elements of the Cordilleran magmatic arc beyond or near the southern terminus of Sevier thrusting imply that proximal facies of the Kaiparowits Formation once extended westward across the Laramide Kaibab uplift toward the Sevier thrust front ∼100 km from the Table Cliff syncline (Fig. 15). A potential means for recycling Kaiparowits sand into younger units during Laramide deformation is to posit removal of Kaiparowits strata from the crest of the Kaibab uplift, which displays 1.6 km of structural relief adjacent to the Table Cliff syncline (Tindall et al., 2010).

Thin remnants of Kaiparowits Formation are exposed locally on the Paunsagunt Plateau near the crest of the Kaibab uplift (Fig. 15), where they conformably overlie middle Cam Wahweap Formation and unconformably underlie the Paleogene Claron Formation (Bowers, 1991), but are not known on the Markagunt Plateau farther west (Eaton et al., 2001; Moore and Straub, 2001; Lawton et al., 2003; Roberts, 2007). On both high plateaus, proximal facies of the capping sandstone member of the Wahweap Formation underlying the Kaiparowits Formation are preserved as the conglomeratic Grand Castle Formation (Lawton et al., 2003; Johnson et al., 2011), which was long regarded as Paleogene in age (Goldstrand, 1990, 1992, 1994; Goldstrand et al. 1993; Goldstrand and Mullett, 1997; Goldstrand and Eaton, 2001) before the recent discovery of Campanian palynomorphs and a dinosaur track in its middle sandstone member (Hunt et al., 2011). The oldest strata lapping across the northern end of the eroded Kaibab uplift toward the relict Sevier thrust front are lacustrine and associated fluvial strata of the Eocene (and possibly Upper Paleocene) Claron Formation (Fig. 2), but stratigraphic relations within the Table Cliff syncline preclude transport of Kaiparowits detritus toward the Uinta Basin from the Kaibab uplift during Paleocene–Eocene Colton sedimentation.

In the Table Cliff syncline, the Kaiparowits Formation is unconformably overlain by the conglomeratic Canaan Peak Formation (80–140 m thick), which was deposited on a coarse alluvial braidplain (Bowers, 1972; Schmitt et al., 1991; Larsen et al., 2010) by eastward paleoflow subparallel to Kaiparowits paleocurrent trends (Fig. 16). Campanian palynomorphs in the Canaan Peak Formation are probably reworked (Eaton, 1991; Goldstrand, 1992, 1994; Goldstrand et al., 1993) because abundant Early Paleocene palynomorphs (Goldstrand, 1990) suggest syntectonic Canaan Peak sedimentation during the post-Kaiparowits Maastricthian–Paleocene time frame of peak Laramide deformation (Fig. 2). Subrounded conglomerate cobbles are dominantly mature quartzite, chert-argillite, and igneous felsite (Bowers, 1972; Schmitt et al., 1991; Goldstrand, 1992). The lack of any sources for those clast types in Kaiparowits or subjacent Cretaceous units of the Kaibab uplift implies delivery of the clasts directly to the Canaan Peak braidplain from the Sevier thrust front or beyond (Schmitt et al., 1991). This observation, coupled with angularity of ∼10° at the Kaiparowits–Canaan Peak unconformity (Bowers, 1972), suggests that Kaiparowits strata had already been stripped from the Kaibab uplift by Maastrichtian or Early Paleocene time. A sandstone sample from the Canaan Peak Formation (Table 1) contains <10% arc-derived detrital zircon grains (Larsen et al., 2010), showing that Canaan Peak sand is not reworked Kaiparowits sand, which contains 50%–90% arc-derived detrital zircon grains.

Overlying the Canaan Peak Formation concordantly along the axis of the Table Cliff syncline (Bowers, 1972; Goldstrand, 1990), but with angularity of 5°–10° (Goldstrand, 1990; Larsen et al., 2010) in the limbs of the syncline, the Pine Hollow Formation (80–120 m thick) is partly equivalent in age to the Colton Formation (Fig. 2) but displays centripetal paleocurrents inward toward the Table Cliff syncline with a weak resultant vector southward along the syncline axis (Fig. 16). The paleocurrrent implication of closed interior drainage for the Pine Hollow Formation within the deforming Table Cliff syncline, bounded by active Kaibab and Circle Cliffs Laramide uplifts (Goldstrand, 1992, 1994; Larsen et al., 2010), is confirmed by Pine Hollow facies patterns. Characteristic cyclic sedimentation encompassed four to six cycles, each fining and drying upward from alluvial fan to playa lake deposits (Larsen et al., 2010) in successions typical of undrained sedimentary basins. Early Paleocene palynomorphs in the basal Pine Hollow Formation (Goldstrand, 1990, 1994; Goldstrand and Eaton, 2001), and Late Paleocene to Early Eocene palynomorphs at higher horizons (Larsen et al., 2010), indicate that no Kaiparowits or other detritus could have exited through the Table Cliff synclinal basin during the time frame of Colton sedimentation in the Uinta Basin (Fig. 2).

The last phase of Paleogene sedimentation recorded by strata in the Table Cliff syncline and across the Kaibab uplift was deposition of lacustrine and associated fluvial strata in the Claron Formation, which formed an Eocene stratigraphic cap (Eaton et al., 2011) over both structural features when Laramide deformation neared completion (Fig. 15). Local incision of paleosols in marginal-lacustrine fluvial deposits record fluctuations in Claron lake level (Goldstrand and Eaton, 2001) that may have reflected waning phases of Laramide deformation.

Detrital zircon populations of post-Kaiparo strata in the Table Cliff syncline (Fig. 18A) reflect minor addition of arc-derived grains (arc indices 7–17) to pre-arc grains either derived from the Sevier thrust belt (Mogollon indices 24–36) or recycled from Cretaceous strata derived ultimately from the Sevier thrust belt (Larsen et al., 2010). Their age spectra resemble middle Campanian (pre-Kaiparowits) Upper Cretaceous strata of the Kaiparowits Plateau and high plateaus to the west (Fig. 18B), and are not markedly different from pre-middle Campanian strata exposed farther east in the Henry Mountains basin (Fig. 1) and nearby areas (Fig. 18C). All have Mogollon indices <45 (Table 3; Fig. 7). The Laramide Table Cliff basin was evidently screened from the California paleoriver by the Circle Cliffs uplift throughout Maastrichtian–Paleogene time (Fig. 1), and lacked any sedimentary connections to wider reaches of the foreland region.

Patterns of detrital zircon ages, paleocurrents, and petrofacies in strata of the Utah foreland suggest that intraforeland uplifts formed during Laramide deformation forced changes in the configuration of fluvial systems carrying sediment longitudinally northward east of the Sevier thrust front. In the Uinta Basin, the pebbly beds present locally as a lateral equivalent of the Tuscher Formation and the Dark Canyon sequence above the Tuscher Formation provide a diachronous record of incipient Laramide deformation, progressing from west to east and in time producing the Uinta Basin, where nearly 5000 m of Paleogene sediment accumulated (Johnson and Johnson, 1991). Before mid-late Campanian time (Fig. 2), data are consistent with master streams flowing along the axis of the Sevier foredeep within ∼100 km of the thrust front and directly over the sites of the younger Kaibab, Circle Cliffs, and San Rafael uplifts (Fig. 1). Beginning in latest Campanian time and continuing into early Paleogene time, the growth of Laramide uplifts blocked sediment transport and shifted longitudinal sediment dispersal through the foreland region to a pathway east of the uplifts.

Growth faults along the East Kaibab monocline flanking the Kaibab uplift bracket the onset of Laramide deformation at 80–76 Ma (Tindall et al., 2010) in middle Campanian time and thinning of Cretaceous strata across the San Rafael Swell farther north document initiation of that uplift as a subsurface growth fold as early as 77 Ma (Aschoff and Steel, 2011) within the same time frame. Initiation of the uplifts as buried subsurface structures did not initially preclude transport of sediment across their evolving crests, but the unconformity between the Kaiparowits Formation and the Canaan Peak Formation in the Table Cliff syncline implies that the Kaibab uplift had attained surficial relief by Maastrichtian time. Ponding of the Maastrichtian to Lower Paleocene North Horn Formation within a depo west of the San Rafael Swell (Lawton, 1983, 1986b) implies the existence of sediment-blocking surficial relief on that uplift during the same time frame. We infer that diversion of fluvial paleoflow in mid-late Campanian time from a Kaiparowits–Farrer pathway across the sites of Laramide uplifts to the younger California paleoriver east of the uplifts for later Tuscher–Colton sedimentation reflected the tectonically induced adjustment of the configuration of the master fluvial systems in the Utah foreland to Laramide deformation. Laramide drainages were forced to thread pathways between Laramide foreland uplifts that grew well to the east of the Sevier thrust front (Fig. 1).

Some detrital zircon populations from Cretaceous–Paleogene strata in northern Arizona are similar to the Colton population (Fig. 19), but others are dissimilar, raising unresolved questions about the configuration of Laramide paleodrainage patterns south of the Utah foreland. In an unsuccessful attempt to intercept Mojave-derived sediment in transit toward the Uinta Basin along an upstream reach of the California paleoriver, we collected two samples (COL12 and COL13; see Table 1) from the Paleogene Music Mountain Formation (Young, 1999) south of the Grand Canyon in northwestern Arizona (Fig. 1). The 2 samples contain only 5% arc-derived detrital zircon grains and none of Paleozoic–Neoproterozoic or Grenville age (subpopulations II and III of Table 3), with 99.5% of the pre-arc detrital zircons derived from Yavapai-Mazatzal and anorogenic granite sources (subpopulations IV and V of Table 3) of the Mogollon highlands (Fig. 19B). Of the 10 arc-derived grains, 2 are Cretaceous (90–99 Ma) and 8 are Jurassic (158–173 Ma), but could derive from Mesozoic plutons intrusive into Yavapai-Mazatzal basement of Arizona rather than from arc assemblages in California. Detrital zircon populations from the Music Mountain Formation resemble those from Oligocene eolianite of the Chuska Sandstone (Figs. 1 and 19A) derived from the deflation of alluvial deposits spread northward from the Mogollon region of central Arizona (Dickinson et al., 2010). Recycling of Music Mountain sand into Chuska sand is not favored, however, by K-S analysis yielding P = 0.01 for comparison of Music Mountain and Chuska detrital zircon populations.

Failure to detect a Colton-like detrital zircon population in the Music Mountain Formation (Fig. 19B) is explicable in any one of four ways, the first two spatial and the latter two temporal. (1) Music Mountain paleodrainages were contiguous northward with paleodrainages on the west side of the Kaibab uplift and did not connect with the California paleoriver flowing east of the Kaibab uplift. (2) Music Mountain paleodrainages were local tributaries to the main stem of the California paleoriver and did not tap Mojave sources. (3) Aggradation of Music Mountain paleodrainages occurred during Laramide deformation when delivery of sediment from the Mojave region to the Uinta Basin was temporarily interrupted during adjustment of drainage patterns to the growth of Laramide uplifts. (4) Music Mountain strata are slightly younger than the Colton Formation and accumulated after the delivery of Mojave detritus to the Uinta Basin.

The Music Mountain Formation filled paleovalleys trending generally from south to north, and also prograded northward as associated alluvial aprons spread across northwestern Arizona south of the modern Grand Canyon (Young, 1999, 2001a, 2001b). Continuation of the Music Mountain paleodrainage system northward across the site of the modern Grand Canyon (Young, 1982, 1985; Graf et al., 1987) would have carried its detritus to the western flank of the Kaibab uplift in Utah (Fig. 15). From that position, Music Mountain sediment could not have crossed the enclosed Paleocene–Eocene (Fig. 2) Pine Hollow depositional system in the Table Cliff syncline to join the California paleoriver farther east, and can be treated as the deposits of a local fluvial system separate from the California paleoriver system. If the Music Mountain paleodrainage instead curved eastward as a local tributary of the California paleoriver system flowing past the southern end of the Kaibab uplift (Fig. 1), its sediment would have enhanced the volume of Mogollon detritus in the pre-arc subpopulations of Colton detrital zircons (Fig. 12A) without contributing non-arc sediment voluminous enough to dilute the arc signature of Mojave-derived detrital zircons in Colton sand.

The Music Mountain Formation is judged from its nonmarine molluscan fauna to be of Early Eocene age (Young and Hartman, 2011; Young et al., 2011), coeval with the Colton Formation, meaning that one of the two spatial explanations for the dichotomy between Music Mountain and Colton detrital zircons is the favored rationale. Other ages for the Music Mountain Formation are considered here only because it has to date yielded no definitive mammalian fauna. If the Music Mountain Formation proves to be slightly older (Paleocene in age), it may have been deposited during the Laramide interval of foreland drainage readjustment when no arc-derived detrital zircons reached the Uinta Basin during North Horn sedimentation (Fig. 4). If the Music Mountain Formation is slightly younger (Middle or Late Eocene), as inferred by Cather et al. (2008), its deposition postdated the delivery of voluminous arc-derived detrital zircons to the Uinta Basin, and the paucity of arc-derived detrital zircons in the Music Mountain Formation presents no conceptual problem.

The age spectrum of detrital zircons in the Upper Cretaceous (younger than 80 Ma) Dome Rock succession (Spencer et al., 2011) of the upper McCoy Mountains Formation (Fig. 19D), exposed near the Colorado River in western Arizona (Fig. 1), yields P = 0.35 from K-S comparison with the Colton age spectrum (Fig. 19C). Arc-derived grains derived from the Mojave segment of the Cordilleran magmatic arc form essentially the same proportion of detrital zircon grains in each case, with the pre-arc grains derived predominantly from subpopulations IV and V (Table 3) of Mogollon provenance. Minor detrital zircon grains in the age range of 250–1250 Ma in the Colton Formation, but absent from the Dome Rock succession, are inferred to reflect admixture of subordinate detritus from the Sevier thrust belt into the California paleoriver during sediment transit northward through the Utah foreland. Dome Rock and Colton detrital zircon populations link both sedimentary successions to sources in the Mojave region, but the two successions are not coeval and provide no information on the evolution of drainage patterns near the Mojave region during intervening Laramide deformation.

The detrital zircon data of Jacobson et al. (2011) for Cretaceous–Paleogene sedimentary assemblages exposed in the southern California forearc allow comparison with data from the Utah foreland backarc to assess the character of arc-derived sediment contributed from the Mojave region to opposite sides of a coastal-inland drainage divide within the Cordilleran magmatic arc (Fig. 1). The forearc successions used for the assessment are those deposited in forearc basins of the western Salinian block, the Sur-Nacimiento block, and the western Transverse Ranges (Fig. 1).

Pre–Middle Paleocene forearc successions (Figs. 20C, 20D, 20F) are dominated by Cretaceous (younger than 120 Ma) grains, suggesting that the drainage divide was then close to the coast, with Jurassic arc assemblages exposed farther east beyond the reach of coastal paleodrainages (Fig. 1). By contrast, backarc successions of Utah received significant components of Jurassic grains (older than 150 Ma) from east of the paleodrainage divide throughout Late Cretaceous time (Fig. 20E, 20G), although comparative data are unavailable for Maastrichtian backarc successions.

For Paleogene successions (Figs. 20A, 20B), Cretaceous and Jurassic subpopulations of arc-derived grains are prominent in both backarc (Colton) and forearc strata, suggesting that the Paleogene drainage divide had migrated far enough inland by Middle Paleocene time for Cretaceous and Jurassic elements of the magmatic arc assemblage to contribute detritus to both forearc and backarc paleodrainages (Fig. 1). Minor differences between the forearc and backarc assemblages are difficult to evaluate, but it is notable that backarc Colton (Fig. 20B) incorporates more of subpopulation Ic (Permian–Triassic) derived mainly from Mexican segments of the magmatic arc (Table 3), and that forearc Eocene strata (Fig. 20A) include an Early Cretaceous subpeak at 117 Ma not present for the backarc. The general similarity of forearc and backarc Paleogene curves is consistent, however, with derivation of arc-derived detritus in the Colton Formation largely from the Mojave segment of the Cordilleran magmatic arc.

Detrital zircon populations derived during Paleogene time from different longitudinal segments of the Cordilleran magmatic arc are similar from the Mojave region northward along the Sierra Nevada (Fig. 21). The age spans of the principal subpopulations document similar ages of igneous source rocks along the full length of the Sierra Nevada–Peninsular Ranges segment of the Cordilleran magmatic arc. The northern, central, and southern Sierra Nevada curves were composited for auriferous gravel samples from the Yuba paleodrainage (Fig. 21A) and the Mokelumne-Stanislaus paleodrainages (Fig. 21B), and from sedimentary assemblages exposed in the El Paso and San Emigdio Mountains (Fig. 21C) aligned along the southern or Tehachapi tail of the Sierra Nevada block.

All the age spectra of Figure 21 display Cretaceous and Jurassic peaks (subpopulations Ia and Ib of Table 3) separated by troughs representing the regional null in arc magmatism. The Sierra Nevada curves (Figs. 21A–21C) display age subpeaks of ca. 115 Ma and 145–150 Ma that are not present for Colton (Fig. 21D) or for southern California forearc assemblages (Figs. 21E, 21F). That observation supports the inference that Colton sand included important contributions of arc-derived grains from the Mojave region far to the south, but not from the Sierra Nevada directly to the west. The presence of an age subpeak at 117 Ma for Eocene forearc assemblages (Fig. 20A) stems from incorporation of samples from farther north on the Salinian block, near the Sierra Nevada, than for the forearc successions (Figs. 21E, 21F) deposited adjacent to conjoined Mojave-Salinia blocks (19 samples with 298 grain ages composited for Fig. 20A but only 16 samples with 196 grain ages for Figs. 21E, 21F).

U-Pb ages of detrital zircons indicate that arkosic sand of the fluviodeltaic Colton Formation in the Laramide Uinta Basin was not derived, as once thought, from Laramide uplifts, but from the Mojave segment of the Cordilleran magmatic arc and associated Yavapai-Mazatzal basement of the Mogollon highlands in southwest Laurentia. Delivery of Mojave-derived detritus to the Utah foreland was accomplished by a hypothetical California paleoriver, which may actually have been an array of related subparallel paleodrainage courses with nearby termini along the southern flank of the Uinta Basin. It is not known where the southern edge of the Uinta Basin was located because the erosional limit of the basin fill as now preserved along the line of the Book Cliffs and Roan Cliffs was not its original depositional limit. No basin-margin facies are exposed along the cliff line.

The course of the California paleoriver from south to north through the Utah foreland was a continuation of longitudinal sediment delivery that had persisted in varying form and volume since initiation of the Cordilleran foreland basin in mid-Cretaceous time. Laramide deformation, however, influenced the configuration of foreland paleodrainages by erecting uplifts as local barriers to fluvial paleoflow. The headwaters of the California paleoriver in the Mojave segment of the Cordilleran magmatic arc (Fig. 1) occupied a paleotopographic syntaxis separating the elevated Nevadaplano (DeCelles, 2004) between the Sierra Nevada segment of the Cordilleran arc and the backarc Sevier thrust belt on the north from comparable uplands to the south, here designated the Mexicoplano (Fig. 1), between the Peninsular Ranges segment of the Cordi arc and the backarc thrust belt of the Sierra Madre Oriental (Lawton et al., 2009). The Nevada and Mexicoplano were comparable in width and geotectonic setting to the modern Altiplano of the Andes (Dickinson, 2011).

The California paleoriver that carried Mojave detritus north to the Uinta Basin south of the Uinta uplift and the sister Idaho paleoriver that later carried Challis detritus south to the Green River Basin north of the Uinta uplift (Chetel et al., 2011) were complementary longitudinal drainages that dominated Laramide sediment dispersal west of the major Laramide uplifts forming the heart of the Laramide province in Colorado and Wyoming. The recognition of these complementary drainages is a major step forward in understanding the paleogeomorphology of the western United States.

Our analysis of Colton provenance and sediment dispersal illustrates how U-Pb ages for detrital zircons can be combined with information about petrofacies and paleocurrents to develop integrated interpretations for the origin of sedimentary assemblages. We show further that the subdivision of detrital zircon populations into constituent subpopulations for statistical analysis can lead to improved understanding of sediment mixing from sources contributing zircons of different ages. Net U-Pb age spectra for individual samples commonly merge age spectra that are signals of different provenances, which served in combination to deliver sediment to depositional sites by different dispersal pathways that shifted over time. Our approach enhances the value of detrital zircons for provenance analysis.

Pat Abbott called our attention to the paleogeographic constraints imposed by Paleogene paleodrainages delineated by him and his colleagues in southernmost California and adjacent Baja California. We appreciate discussions of Laramide lacustrine sedimentation with Alan Carroll. We thank Elizabeth Cassel, Robinson Cecil, Gary Hunt, Carl Jacobson, and Paul Link for providing Excel files of U-Pb ages for their detrital zircon samples, and Tom Fouch for a detailed measured section of the strata exposed in Gate Canyon. Paul Heller provided data from the unpublished thesis of Genevive Mathers. Brady Foreman discussed with us his unpublished paleocurrent data for the DeBeque Formation. Carol A. Hill provided guidance to Duff Brown Tank (sample COL13). Comments by Jon Spencer improved the text. Jacqueline Dickinson helped collect detrital zircon samples. Jim Abbott prepared all the figures. Comments from Carol Frost, Chris Fedo, and two anonymous reviewers sharpened the text.