Post-Laramide, Eocene epeirogeny in central Colorado—The result of a mantle drip?

Today's Southern Rocky Mountains (SRM) consist of multiple discrete ranges and flanking synorogenic basins (Fig. 1) that were produced by crustal shortening during the Laramide orogeny (ca. 75–45 Ma; Tweto, 1975; Dickinson et al., 1988). Although the modern mountains are built on this orogenic foundation, broad consensus exists that the SRM achieved the current surface elevation (sensu England and Molnar, 1990) of ~2.8 km due to epeirogenic uplift (e.g., Trimble, 1980; Bird, 1988; Eaton, 2008; Cather et al., 2012; Karlstrom et al., 2012).

Consensus evaporates when questions turn to when and why the SRM and adjacent Great Plains experienced epeirogenic uplift. Modern SRM ranges are exhumed facsimiles of their Laramide predecessors. Post-Laramide deposition switched to erosion ca. 5 Ma, stripping easily eroded sediment from the resistant, previously buried crystalline cores to re-expose the Laramide ranges. Many workers thus conclude epeirogeny occurred ca. 5 Ma (e.g., Epis and Chapin, 1975; Trimble, 1980; Karlstrom et al., 2012). Others argue global cooling at 5 Ma caused the switch to erosion; they favor earlier epeirogeny (e.g., Molnar and England, 1990; Gregory and Chase, 1992; Pelletier, 2009). Some workers suggest an Eocene or Oligocene start to a still active epeirogeny (e.g., Sahagian et al., 2002; Eaton, 2008). Others invoke multiple epeirogenic episodes with different causes (e.g., Cather et al., 2012; Karlstrom et al., 2012). This surfeit of hypotheses highlights the need to constrain the timing of post-Laramide surface uplift event(s) to preclude some mechanisms and elevate others to a short list of likely candidates.

We present new low-temperature thermochronology for central Colorado's Arkansas Hills that documents Eocene exhumation beginning between 54 and 46 Ma, long after Laramide crustal shortening had ended. Next, we examine the sedimentary histories of the South Park–High Park and Denver basins (Fig. 1), showing that Arkansas Hills exhumation was coeval with formation of basin unconformities that mark the end of subsidence there. We interpret these coeval events to record Eocene epeirogenic surface uplift in central Colorado. Lastly, we explore published mechanisms to explain the epeirogeny and suggest a new one—foundering of a lithospheric drip.

We distinguish here between epeirogenic and orogenic surface uplift, so we must establish the temporal, spatial, and process connotations of our use of the term “Laramide orogeny,” a term that has been ascribed multiple meanings. Here we follow Tweto (1975) and DeCelles (2004), defining the Laramide orogeny as an episode of basement-involved reverse faulting in the Cordilleran foreland between Late Campanian and middle Eocene time.

Given this regional ca. 75–45 Ma timing for the Laramide orogeny, studies commonly label any rock uplift or exhumation event that transpired between 75 and 45 Ma as “Laramide” and ascribe its cause to crustal shortening. But the initiation and cessation of crustal shortening (i.e., orogenic mountain building) was diachronous across the Laramide province (e.g., Tweto, 1975; Copeland et al., 2017). A given SRM event that falls within the canonical “Laramide” time window may or may not have been caused by crustal shortening. Here we use the term “Laramide” to refer to a crustal shortening (i.e., orogenic) event or its timing in a specific SRM range. We classify any event that postdates crustal shortening in that range as “post-Laramide,” even if it occurred during the 75–45 Ma time span of regional Laramide orogenic activity. As argued below, Laramide shortening ended in our Arkansas Hills study area by 67 Ma; any post–67 Ma event, including the epeirogenic surface uplift we infer, is therefore “post-Laramide.”

Our study area, which stretches from the Arkansas Hills to High Park (Fig. 1), has experienced a complex geologic history. Here we recap that history, highlighting events of particular relevance for assessing the implications of our thermochronologic results (Table 1).

Central Colorado's Proterozoic crystalline basement underlies a 300-m-thick early Paleozoic sedimentary sequence (Tweto, 1987). The Pennsylvanian–Permianaged Ancestral Rockies orogeny then raised two highlands separated by a sedimentary basin—the Central Colorado Trough (Ye et al., 1996; Barkmann et al., 2016). Older sediments were stripped from the highlands while kilometers of synorogenic sediment accumulated in the Central Colorado Trough. The Arkansas Hills (Fig. 1; Table 1), which lay in the ancient basin, contain 3.4–5.8 km of Paleozoic sediment (Wofford, 1986). High Park lies in the ancient highlands, so Jurassic sediments unconformably overlie Precambrian basement (Figs. 2 and 3; Wobus et al., 1979).

Late Mesozoic Colorado was covered by the Western Interior Seaway (WIS), in which 1.5–3.0 km of sediment accumulated (Cross and Pilger, 1978), with 500–1600 m now exposed in the Arkansas Hills (Wofford, 1986). When the Laramide orogeny began, Arkansas Hills basement was mantled by 5–8 km of Phanerozoic sediment; whereas in High Park, the basement had only 0.5–1.5 km of sedimentary cover (Fig. 2; Table 1).

The asymmetric Sawatch anticline rose early in the Laramide orogeny (Tweto, 1975) atop the Elk Range thrust and the Castle Creek reverse fault, both near Aspen (Fig. 1; Table 1; Bryant, 1966; Kellogg, 1999). It rose above sea level after 75 Ma based on Mancos Formation ammonite dates; faulting was active by 72–70 Ma (Tweto, 1975). Later rifting split the anticline (see Development of the Rio Grande Rift [RGR] section), separating the Arkansas Hills, its southeastern flank, from the Sawatch Range (Fig. 4A).

The Arkansas Hills consists primarily of eastdipping Paleozoic sediment overlying basement. The syn-Laramide Whitehorn Granodiorite cross-cuts the Weston fault (Fig. 3; Wrucke, 1974; Wrucke and Dings, 1979; Wallace et al., 2000), which is the dominant Laramide structure in the Hills. Thus, Arkansas Hills thrusting ended before Whitehorn intrusion. Abbey et al. (2017) obtained a 67 Ma zircon U/Pb date on the Whitehorn (67.31 +0.57/−0.78 Ma). Wofford (1986) constrained pluton emplacement to 3.0–7.5 km based on sediment thickness; Abbey et al. (2017) narrowed that to 5–7 km depth (see below). Farther north, in the Mosquito Range, fault activity continued into the early Paleocene (post–65–63 Ma; Bohannon and Ruleman, 2013).

Sedimentary histories in the Piceance Basin to the west (Obradovich, 1969; Foreman and Rasmussen, 2016) and South Park to the east (Fig. 1; Tweto, 1975; Barkmann et al., 2016) confirm a late Cretaceous–early Paleocene uplift age for the Sawatch anticline. Paleocurrents and provenance (i.e., Paleozoic clasts) in early synorogenic sediment of both basins have a Sawatch source. The 69 Ma base of the South Park Formation (South Park's synorogenic fill) rests unconformably atop progressively younger rocks farther east, revealing pre–69 Ma Sawatch uplift and erosion (Barkmann et al., 2016). Clast provenance and east-flowing paleocurrents continue to record a Sawatch source until 60 Ma, based on the 59.7 ± 2.0 Ma K/Ar date on a tuff just below an intraformational unconformity (Bryant et al., 1981; Barkmann et al., 2016).

The upper South Park Formation then records an eastward shift in the locus of shortening from the Sawatch to the Front Range. Paleocurrents reverse to a westward flow, sourced from the Front Range, above the unconformity. Paleozoic clasts disappear in the upper, arkosic, 60–56 Ma portion of the formation, replaced by clasts of Precambrian granite and gneiss (Barkmann et al., 2016) derived from the Front Range. The South Park Formation itself was then underthrust beneath Precambrian basement along the west-vergent Elkhorn Thrust during Laramide rise of the Front Range (Obradovich et al., 1969; Tweto, 1975; Raynolds, 1997; Barkmann et al., 2016).

To summarize, although the Laramide orogeny continued elsewhere in the Rockies until ca. 45 Ma, crustal shortening and concomitant rock uplift ceased in the southeastern Sawatch anticline, the location of our study area, by ca. 67 Ma (Table 1). Exhumation of the Sawatch highland in response to that crustal shortening had dwindled by 60 Ma.

Extensive erosion in the aftermath of the Laramide orogeny planed off the Laramide Rockies to form the widespread, mostly low relief Rocky Mountain Erosion Surface (RMES) before 38 Ma (Fig. 2; e.g., Epis and Chapin, 1975; Scott, 1975; Colman, 1985; Scott and Taylor, 1986). We know the RMES in central Colorado had formed by the late Eocene because it is blanketed by ash flows erupted during regionally extensive volcanic activity known as the ignimbrite flare-up (e.g., Epis and Chapin, 1974; Farmer et al., 2008). Lipman (2021) argued the Sawatch region's paleoelevation was high based on deep, rapid erosion of its flare-up volcanoes.

The late Eocene Mosquito Range and Arkansas Hills were cut by three major east-flowing paleovalleys that filled with sediments and ash-flow tuffs as old as 38 Ma (McIntosh and Chapin, 2004). The three paleovalleys are, from north to south, the Buffalo Peaks, Trout Creek, and Salida–Waugh Mountain (Figs. 1 and 3). The Salida-Waugh Mountain valley cuts the Whitehorn Grano diorite (Figs. 2, 3, and 4A); the valley possessed of >450 m of late Eocene topographic relief to the top of Cameron Mountain, composed of that granodiorite (Fig. 4A; Wallace et al., 1997; Wallace and Lawson, 2008).

An early, extensive flare-up ash flow is the 36.7 Ma Wall Mountain Tuff (WMT), whose source lies in the Sawatch Range near the Mount Princeton batholith (Fig. 1; McIntosh and Chapin, 2004). The WMT flowed east through all the paleovalleys, across the RMES, and spilled onto the Great Plains at Castle Rock (Figs. 1 and 2), 120 km to the east. The WMT is not significantly displaced across the Rockies–Great Plains interface, so post–37 Ma differential uplift between the provinces has been minimal (Leonard and Langford, 1994). Several studies have combined the province-spanning nature of the WMT with paleofloral characteristics in the 34.1 Ma Florissant Formation immediately above to deduce paleoelevations, and hence the timing of epeirogenic uplift, in central Colorado. Unfortunately, their paleoelevation estimates do not agree. Meyer (1992), Gregory and Chase (1992) and Wolfe et al. (1998) concluded the RMES at Florissant (Fig. 1) stood at its current 2.5 km elevation by 34 Ma, meaning epeirogenic uplift must be late Eocene or older. Other workers, using similar data but different techniques, concluded that Florissant stood then at 300–900 m; epeirogenic uplift elevated it to its current height more recently (MacGinitie, 1953; Trimble, 1980; Cather et al., 2012; Zaborac-Reed and Leopold, 2016). Clearly, other methods must be employed to deduce epeirogenic uplift timing.

Colorado's stress field evolved from compression to extension during the Oligocene. Normal faults began to dismember the RMES by 28 Ma (e.g., Scott, 1975; Landman and Flowers, 2013). Several Laramide ranges were bisected by RGR half-grabens (Kellogg, 1999). The Sawatch anticline was split by growth of the Upper Arkansas Valley half-graben, separating the Arkansas Hills from the rest of the anticline (Figs. 1 and 4A). Two thermo-chronologic studies in the Sawatch Range tracked exhumation associated with rifting (Ricketts et al., 2016; Abbey and Niemi, 2018), but the large magnitude of the RGR signal there obscures the pre-RGR history, which is our focus. The Arkansas Hills, in the half-graben's hangingwall, did not experience RGR-associated exhumation, making them an ideal place to deduce that earlier history.

The apatite, zircon and titanite (U-Th)/He (AHe, ZHe, and THe) techniques and apatite fission-track (AFT) method are thermochronologic tools to resolve thermal histories for rocks below ~200 °C. These thermal histories can be used to infer exhumation histories through the upper ~5–7 km of the crust based on assumptions about the upper-crustal thermal field.

The (U-Th)/He method is based on the decay of U, Th, and Sm to He, and on volume diffusion of He out of the crystal lattice. At low temperatures, the crystal retains all He; at high temperatures, it is completely lost; and at intermediate temperatures, He is partially retained in what is known as the He partial retention zone (PRZ; Fig. 5). The temperature range of He retention depends on the mineral, its amount of radiation damage (Shuster et al., 2006), and on crystal size (Reiners and Farley, 2001). Apatite is sensitive to ~30–60 °C for low-damage crystals, increasing to ~110 °C for high-damage apatite (Farley, 2000; Flowers et al., 2009). For zircon and titanite, low-damage crystals begin to retain He at ~210 °C (Fig. 5; e.g., Reiners and Farley, 1999; Reiners et al., 2002; Stockli and Farley, 2004). Damage accumulation first increases temperature sensitivity in zircon; then, at higher levels, decreases it (in both zircon and titanite) to <50 °C (Guenthner et al., 2013; Baughman et al., 2017).

Effective uranium concentration (eU) is used as a proxy for radiation damage for grains of the same mineral that underwent the same thermal history (eU weighs the decay of U and Th by their He production, and is computed as eU = U + 0.235 * Th). Depending on the time-temperature (tT) path and the eU range of the mineral suite, positive (apatite and zircon), negative (zircon and titanite), or flat (apatite, zircon, and titanite) date-eU relationships occur (e.g., Shuster et al., 2006; Flowers et al., 2007; Guenthner et al., 2013). He diffusion kinetic models that track the evolution of mineral He retentivity with radiation damage accumulation and annealing are available for apatite (e.g., Flowers et al., 2009; Gautheron et al., 2009) and zircon (e.g., Guenthner et al., 2013), and can be used to quantitatively model the thermal history significance of (U-Th)/He results (e.g., Ketcham, 2005; Gallagher, 2012).

The basis of the apatite fission-track method is the accumulation of fission tracks in the crystal lattice from spontaneous nuclear fission of ²³⁸U, and on temperature-dependent annealing of those tracks. At temperatures above the partial annealing zone (PAZ; Fig. 5), ~120 °C, fission tracks anneal immediately. Within the PAZ, down to ~60 °C, tracks are progressively shortened (partially annealed; from initially ~16 µm length). At temperatures below the PAZ, accumulated tracks are preserved (e.g., Fleischer et al., 1975; Gleadow and Fitzgerald, 1987; Donelick et al., 1990). Track annealing rates depend primarily on the temperature but also on factors such as heating and/or cooling rate, chemistry (in apatites), track orientation, and radiation damage. A kinetic parameter, the mean of a grain's maximum diameter of fission-track etch pits (D_par), can indicate how readily the tracks anneal (e.g., Carlson et al., 1999). Determining spontaneous fission-track density and the mineral U content allows calculation of the time since track accumulation started. The distribution of confined track lengths indicates the integrated thermal history of the mineral (e.g., review by Donelick et al., 2005, and references therein).

A previous exhumation study in the Arkansas Hills included zircon U/Pb, hornblende and biotite ⁴⁰Ar/³⁹Ar, and AHe analyses of the Whitehorn Granodiorite and AHe and ZHe analyses of nearby Precambrian samples (Abbey et al., 2017). The overlap at ca. 67 Ma in zircon U/Pb and all ⁴⁰Ar/³⁹Ar dates was used to infer Whitehorn pluton emplacement and rapid cooling to T <300 °C at that time. These results and existing stratigraphic estimates (Wofford, 1986) imply that the Whitehorn pluton was emplaced at 5–7 km depth at 67 Ma, then subse-quently exhumed (Abbey et al., 2017). In contrast, pre–67 Ma AHe dates for lower Arkansas River samples (east of Coaldale; Fig. 1) indicated less exhumation to the east. Whitehorn Granodiorite AHe dates in the Abbey et al. (2017) study were not fully reproducible. This led the authors to disregard these data when interpreting the exhumation history of the Arkansas Hills and to instead attribute sample age dispersion to partial resetting from passage of ignimbrites through the paleo valleys.

Three of our four Arkansas Hills samples are Whitehorn Granodiorite. The samples span the 447 m elevation range of Whitehorn exposure, from the Salida–Waugh Mountain paleovalley (2899 m) to the top of Cameron Mountain (3346 m; Figs. 2, 3, and 4A). The fourth sample (2688 m) is Precambrian granodiorite from the Gribbles Run paleovalley (Figs. 2 and 3). All three High Park samples are Precambrian granitoids collected at elevations between 2433 and 2508 m (Figs. 2 and 3).

The Whitehorn Granodiorite presents two advantages for our study. First, its Cretaceous age means its apatite crystals have accumulated less radiation damage than Proterozoic rocks, thus they have lower He retentivity. Consequently, Whitehorn AHe should detect lower exhumation magnitudes than will basement samples. Second, the White-horn's thermal and exhumation history is shorter and simpler than that of Proterozoic basement, making its interpretation less complex. See Stanley et al. (2013, 2015) for an example of this approach in southern Africa.

All our samples either directly underlie the 37 Ma Wall Mountain Tuff or are from Cameron Mountain, which is cut by the tuff-filled Salida–Waugh Mountain paleovalley (Figs. 3 and 4A). Therefore, every sample was on Earth's surface by 37 Ma. This knowledge greatly restricts the time interval during which any thermochronologically detected exhumation event could have transpired.

Mineral separation was done at the University of Colorado using standard crushing, water table, magnetic, and heavy-liquid techniques. Single crystals of zircon, titanite, and apatite were then hand-picked for (U-Th)/He analysis based on crystal size, shape, and absence of inclusions using a microscope with polarized transmitted and reflected light. All (U-Th)/He analyses were done in the University of Colorado Thermochronology Research and Instrumentation Lab (CU TRaIL). Detailed analytical methods are described in the Supplemental Material¹.

Sample mean (U-Th)/He data are reported in Table 2, (U-Th)/He single-grain data in Table 3, and additional data details in Table S1 (^{footnote 1}). Analytical uncertainties on single-grain dates are reported at 2σ and include propagated analytical uncertainties on the parent and daughter isotope measurements. We assign a conservative 15% uncertainty on single-crystal eU values based on past work (e.g., Guenthner et al., 2013; Baughman et al., 2017). Where sample means are noted in the text and in Table 2, the associated uncertainty is reported as the 1σ sample standard deviation.

Fission-track dating of apatites was done at the University of Tübingen using the external detector and zeta-calibration techniques (Hurford, 1990). Apatites were mounted in epoxy resin, ground, polished, etched (for 20 s in 5.5 M HNO₃ at 21°C), and covered with mica external detectors for irradiation. Samples were irradiated, together with Durango and Fish Canyon Tuff age standards and IRMM540 uranium dosimeter glasses, at the FRM-II nuclear reactor (Garching, Germany). After irradiation, micas were etched in 40% HF for 30 min at room temperature. Fission tracks were counted, and confined tracks and D_par were measured at 1000× power using a Zeiss AxioImager microscope with AutoScan software. Five D_par measurements were conducted for each grain, where track density was known and/or confined track lengths were measured. Table 4 is a summary of AFT data, including AFT pooled ages, mean confined track lengths, and D_par. Single-grain data used in thermal history modeling are in Tables S2 and S3 (^{footnote 1}).

To accomplish our goal of reconstructing the Whitehorn pluton's cooling and erosion history between its 67 Ma time of emplacement and its surface exposure by 37 Ma, we acquired ten ZHe dates and 20 AHe dates from all three Whitehorn samples, six THe dates from one sample, and AFT dates and track-lengths for two samples (Tables 2–4, S1–S3, ^{footnote 1}).

The individual ZHe dates range from 69 to 55 Ma, overlapping with or slightly younger than pluton emplacement age. Dates show no correlation with eU (55–337 ppm; Fig. 6A). Samples yield mean ZHe dates of 65 ± 3 Ma, 59 ± 1 Ma, and 59 ± 5 Ma for BF16-1, LA17-2, and LA17-1, respectively. Six single-grain THe dates from BF16-1 are 72–61 Ma (eU = 49–102 ppm). The mean THe date of 65 ± 4 Ma overlaps with the sample's ZHe date.

We acquired AFT data for two samples, but only one (LA17-2) yields a reliable result. The AFT date for LA17-2 (24 measured grains) is 63 ± 4 Ma (1σ; Fig. 6A; Tables 4 and S2, ^{footnote 1}) and overlaps with pluton emplacement. The mean track length is 13.6 ± 2.1 µm (1σ; Table 4) with a unimodal distribution (Table S3). The latter indicates that no reheating occurred once the sample cooled below PAZ temperatures. While the rather long mean track length suggests that the sample did not remain within the PAZ for an extended length of time, the relatively wide track length distribution also indicates that cooling did not occur rapidly. Note however that only 27 confined track lengths could be measured (not the typical goal of 100) and that the sample's thermal history is not well constrained from these results. Single-grain D_par range between 2.9 and 5.3 µm, and single-grain dates do not correlate with D_par (Table S2). The relatively high mean D_par of 3.8 ± 0.6 µm (Table 4) suggests comparatively high resistance to track annealing and, thus, a high closure temperature (e.g., Carlson et al., 1999; Donelick et al., 2005). The AFT date for BF16-1 is 79 ± 4 Ma from 20 measured grains (Tables 4 and S2, ^{footnote 1}) and is significantly older than the 67 Ma U/Pb emplacement age. Many grains showed curved dislocations in the crystal lattice when etched. Based on the pluton crystallization age, we consider the AFT data for sample BF16-1 unreliable and do not use them in the subsequent analysis.

The 20 single-grain AHe dates from the three samples are all distinctly younger than pluton emplacement and vary from 55 to 41 Ma. The dates are uncorrelated with eU (13–43 ppm eU range; Fig. 6B), crystal size (Fig. S1A, ^{footnote 1}), or sample elevation (Fig. 6C). Sample mean dates and uncertainties overlap and are 53 ± 2 Ma, 47 ± 4 Ma, and 47 ± 4 Ma for LA17-1, LA17-2, and BF16-1, respectively. Our AHe data are reproducible both within and between all samples. This contrasts with some AHe data reported previously for the Whitehorn Granodiorite (Abbey et al., 2017), as discussed in the Comparison with Previous Work on the Whitehorn Granodiorite section of the text file in Supplemental Materials. Based on the reproducibility of our AHe results, we consider them reliable; they faithfully record the cooling and exhumation history of the constituent rocks.

We acquired ZHe and AHe data for a 1.67 Ga Arkansas Hills granodiorite, FN17-1, collected 10 km east and 211 m lower than LA17-1 to compare with data from the nearby Whitehorn pluton. Three ZHe grain dates are reproducible (64–57 Ma) and uncor-related with eU (793–961 ppm, Fig. 7A). The mean 61 ± 4 Ma date overlaps the Whitehorn ZHe dates. A high eU grain (1747 ppm) has a 39 Ma ZHe date, likely due to radiation-damage enhanced loss of He retentivity (Fig. 5; Guenthner et al., 2013). Four AHe dates are 70–59 Ma for a 46–166 ppm eU range (Fig. 7A). The 63 ± 5 Ma mean AHe date overlaps sample FN17-1's ZHe date and is older than AHe dates for the nearby Whitehorn samples.

The 18 AHe dates for three Precambrian High Park samples (Figs. 3 and 6B; Tables 2 and 3) allow assessment of spatial heterogeneity of the regional thermal history. The AHe dates are almost all pre-Laramide, and intrasample variability is high. Samples have positive date-eU correlations (Fig. 7B), consistent with the expected increase in He retentivity with greater radiation damage (Flowers et al., 2009) but have no correlation with crystal size (Fig. S1B, ^{footnote 1}). Dates are 653–135 Ma (N = 7) for FN17-2, 173–59 Ma (N = 6) for FN17-3, and 520–370 Ma (N = 5) for FN17-4.

Our primary goal is to constrain the post-Laramide cooling and erosion history of the study area. Because the High Park samples were incompletely reset in the Mesozoic, they cannot resolve that history. For that reason, we briefly discuss the High Park data and then focus on the implications of the Arkansas Hills samples for post-Laramide exhumation.

The pre–Western Interior Seaway (WIS) AHe dates for High Park samples limit maximum temperatures to <110 °C since Laramide time, meaning burial depths of <~2.5–3 km. By contrast, Arkansas Hills ZHe and THe dates are Laramide or younger, so temperatures were >180 °C (burial >5 km) when the Laramide orogeny began. This east-to-west depth increase reinforces Abbey et al.'s (2017) conclusion that Cenozoic exhumation is greater west of Coaldale (Fig. 1).

The cumulative WIS and Laramide sedimentary thickness in South Park, 70 km north of High Park exceeds 4.5 km (Barkmann et al., 2016). Thus, the maximum 2.5–3 km post-WIS basement burial depth for our High Park samples requires significant southward thinning of WIS and/or Laramide synorogenic sedimentary strata, consistent with Cross and Pilger's (1978) WIS isopach map.

In contrast to the High Park samples, our Arkansas Hills data enable detailed interpretation of the Laramide and younger thermal and exhumation history. Comparison of the multiple thermochronometer results for the Whitehorn samples implies a multiphase cooling and erosion history in the 30 m.y. between 67 Ma pluton emplacement and surface exposure by 37 Ma. The general overlap of ZHe, THe, and AFT dates with crystallization age indicates an initial episode of rapid Paleocene cooling from magmatic temperatures down to temperatures less than ~120 °C based on the AFT results. Eocene AHe dates for all samples suggest a second cooling phase, from ~120 °C to the surface, during the Eocene.

The Arkansas Hills Precambrian basement sample (FN17-1), collected 10 km east of the Whitehorn samples (Fig. 3), likely has a similar thermal history, but its AHe date (63 ± 5 Ma) is older. This disparity in AHe dates is likely because the sample's Precam-brian apatite grains have higher He retentivity (i.e., higher closure temperature) than the Whitehorn apatite due to a longer interval of radiation damage accumulation (Fig. 5; Flowers et al., 2009).

We conducted thermal history modeling to quantitatively evaluate evidence for an Eocene cooling signal in the Arkansas Hills samples (Fig. 8). Our inverse time-temperature (tT) thermal history models use the HeFTy software (Ketcham, 2005), RDAAM annealing model (Flowers et al., 2009) for apatite He diffusion kinetics, ZRDAAM annealing model (Guenthner et al., 2013) for zircon He diffusion kinetics, and the model of Ketcham et al. (2007) for apatite fission-track annealing kinetics. We do not include THe data because there is no titanite radiation damage model, but the overlap of our THe and ZHe dates means their inclusion would not impact model results. HeFTy supports a maximum of seven date inputs; we grouped the AHe and ZHe dates into bins of similar eU and used the average date, grain size, and eU of each bin as a separate input (Figs. 5A and 6A), which is a common HeFTy modeling strategy. AFT date and track length form another input. HeFTy tests the predicted dates for thousands of tT paths against the input data. Those paths that surpass a minimum goodness-of-fit threshold are deemed “good”—or “acceptable”—fits. Full model details are in Table S4 (^{footnote 1}).

Figure 8 shows thermal history model results for ZHe, AFT, and AHe data for all four Arkansas Hills samples. Geologic constraints that we apply to all Whitehorn models (Figs. 8A–8C) are: (1) temperature (T) = 300 °C at 67 Ma (pluton emplacement); (2) cooling to surface T = 0–20 °C by 37 Ma (Wall Mountain Tuff contact); and (3) surface temperatures today. We add a T = 0–40 °C exploration box from 65 to 39 Ma to force HeFTy to test for Paleocene cooling to near-surface temperatures (Figs. 8A and 8B reveal that such early cooling is impossible for BF16-1 and LA17-2).

Figure 8D shows the last 100 m.y. of the Precambrian basement sample model (FN17-1). See Figure S3 (^{footnote 1}) for the full 1680 Ma model. Geologic constraints we apply to this model (see Wallace et al., 1999) are: (1) T = 300 °C at t = 1672 Ma (pluton emplacement age); (2) cooling to T = 0–20 °C by t = 490–445 Ma (the age of the unconformably overlying Manitou Limestone); (3) return to surface T = 0–20 °C by t = 37 Ma (age of the overlying Wall Mountain Tuff); (4) surface T = 0–20 °C today. FN17-1 was buried by 5–8 km of Phanerozoic sediments after exposure on the surface in the Ordovician, so we apply an exploration box of T = 0–200 °C from t = 445–80 Ma to allow for reheating.

All tT paths with a goodness of fit (GoF) above 0.5 (labeled “good” paths in HeFTy) are shown in orange on Figure 8; the “best-fit” path is black. Paths with 0.05<GoF<0.5 (HeFTy's “acceptable” fits) are shown in progressively warmer colors for increasing GoF.

Do the quantitative tT models validate the above qualitative interpretation that the Arkansas Hills experienced two episodes of rapid cooling separated by an interval of slower cooling? White-horn samples BF16-1 and LA17-2 (Figs. 8A and 8B) place the tightest constraints on the cooling history—BF16-1 because its apatite and zircon grains span a wide eU range and LA17-2 because of its additional AFT constraint. In both cases, because HeFTy must honor data at multiple closure temperatures, the range of permissible paths is limited. Scrutiny of both tT models confirms the pluton's two-phase cooling history. The less constrained LA17-1 and FN17-1 (Figs. 8C and 8D) models are consistent; they permit but do not require twostage cooling during the same interval.

For samples LA17-2 and BF16-1, the good-fit (orange) tT paths (Fig. 8) do not cool to <40 °C until after 54 Ma, thus yielding the two-phase cooling history and requiring substantial cooling between 54 Ma and surface conditions at 37 Ma. This significant phase of post–54 Ma cooling dictates the sinuous shapes of the orange fields in Figures 8A and 8B. Figure 9 shows that the best-fit paths for both samples predict grain dates that agree with measured dates, demonstrating the reliability of both models.

The orange field for BF16-1 contains 354 separate good paths, and LA17-2 has 618 of them (out of 10,000 paths attempted; Figs. 8A and 8B), making it impossible to trace individual paths across the entire tT history. Each good path represents an equally plausible tT history, so our two-stage cooling interpretation requires demonstration that all good paths exhibit two-stage cooling, not just some of them. To facilitate the tracing of complete paths, we plotted 40 randomly selected good paths for BF16-1 in gray on Figure 10A. Every path defines a two-stage cooling history, with one rapid cooling episode in the Paleocene and one in the Eocene. Figure 10B shows the same result for 40 different randomly selected BF16-1 paths. Figures 10C and 10D show different randomly selected paths for LA17-2—all display two-stage cooling. The evidence for two-stage cooling is robust.

Tracking the field of good fits shows that the pluton cooled to between 120 and 60 °C by 60 Ma when cooling slowed. Part of this cooling must be due to postemplacement thermal relaxation, but 120–60 °C is unreasonably cool for the 5–7 km emplacement depth, so Laramide exhumation must also play a role. Assuming today's 30 °C/km geothermal gradient (Berkman and Watterson, 2010) and surface T = 10–20 °C, the pluton lay ~1.3–3.7 km below the surface at 60 Ma. Thus, the Arkansas Hills experienced ~1.3–5.7 km of exhumation between 67 and 60 Ma. This thermal modeling result showing that rapid Laramide erosion ended by 60 Ma is consistent with sedimentary evidence that the Sawatch anticline stopped supplying sediment to South Park at that time (Barkmann et al., 2016).

After an episode of comparatively slow cooling between ca. 60–54 Ma, a second phase of rapid cooling began sometime between 54 and 46 Ma; the pluton was exhumed from ~1.3–3.7 km depth at the start of this episode and reached the surface by 37 Ma. Laramide crustal shortening was not responsible for this Eocene exhumation because the locus of shortening had moved east of the Sawatch anticline by that time. We attribute it, instead, to epeirogeny.

The onset of epeirogenic rock uplift could trigger the kilometer-scale, Eocene Arkansas Hills exhumation documented by the tT models, but thermochronology alone cannot constrain the duration of the epeirogeny, and our data pertain to one small area—epeirogeny, by definition, affects a broad region. The sedimentary histories of the South Park–High Park and Denver basins (Fig. 1) provide four strong arguments for the simultaneous onset of Eocene erosion in those basins and the Arkansas Hills, a hallmark of epeirogeny, and that the epeirogeny affected central Colorado sedimentation patterns into the Oligocene. These four arguments are: (1) development of basin-wide unconformities simultaneous with onset of Arkansas Hills exhumation; (2) subsequent deposition of late Eocene conglomerate; (3) cut-and-fill architecture in subsequent deposits; and (4) missing synorogenic strata in the Laramide basins. Here we examine each argument.

Accommodation space was created by subsidence of the South Park–High Park Basin throughout deposition of the 69–56 Ma South Park Formation, the basin's synorogenic fill (Barkmann et al., 2016). In South Park, an unconformity separates the South Park Formation from 38.2 Ma tuff (McIntosh and Chapin, 2004); therefore, the unconformity spans 56–38 Ma. Any South Park Formation deposited in High Park was eroded during this unconformity (Figs. 2 and 4B).

The Denver Basin exhibits an unconformity that spans the same time interval. The basin's Laramide synorogenic fill (Figs. 2 and 4C) has a confusing history of name changes; most recently, Thorson (2011) called it the Denver Basin Group. It is ~900 m thick and was deposited between ca. 69–54 Ma (Hicks et al., 2002). An unconformity internal to the Group exists on the basin edges, but the missing middle-late Paleocene strata exist in the basin's center (Thorson, 2011), revealing that subsidence produced accommodation space continuously during the orogeny until 54 Ma, the age of the upper Dawson Arkose, the Group's stratigraphically highest formation.

Near Castle Rock, the Denver Basin Group fills the basin almost to the level of the RMES on the adjacent Front Range (Figs. 1 and 2). The oldest unit overlying the Dawson Arkose is the Larkspur Conglomerate (Thorson, 2011). Detrital zircon ages reveal the Larkspur to be younger than 41 Ma (Koch et al., 2018), and it underlies the 37 Ma Wall Mountain Tuff. Thus, the Denver Basin unconformity spans the time range of 54 Ma to 41–37 Ma.

In summary, unconformities in South Park (56–38 Ma) and the Denver Basin (54–41 Ma) are contemporaneous with Arkansas Hills exhumation (Table 1), demonstrating that post-Laramide exhumation affected both Laramide highlands and basins, a hallmark of epeirogeny.

When deposition resumed at 41–37 Ma in the Denver Basin, the sedimentary system's energy level was considerably higher than it was at the end of Laramide deposition. The maximum grain size in the Dawson Arkose is 1.5 cm; the overlying Larkspur Conglomerate contains rounded granitic clasts to 20 cm (Thorson, 2011). The Larkspur contains few volcanic clasts (Koch et al., 2018), and it underlies the Wall Mountain Tuff, the first extensive ignimbrite flare-up deposit. Because this rise in transport energy manifests in the pre-volcanic Larkspur Conglomerate, the rise of ignimbrite flare-up volcanic topography could not be its cause.

Although the names of clastic units differ in South Park–High Park, their characteristics mirror those of the Denver Basin, documenting that late Eocene high-energy conditions predated the flare-up. The oldest sedimentary unit overlying the South Park Formation is the Echo Park Alluvium, of likely late Eocene age (Epis and Chapin, 1974). It fills paleochannels and two small grabens (Figs. 2 and 3; Epis and Chapin, 1974; Wobus and Epis, 1978; Epis et al., 1979). It is an analog for the Larkspur Conglomerate in that it largely lacks volcanic clasts and underlies the Wall Mountain Tuff.

Many workers concluded that Colorado's late Eocene topography possessed low relief (e.g., Epis and Chapin, 1975), but ubiquitous low relief is incompatible with widespread deposition of late Eocene conglomerate (Evanoff, 1990). As noted above, Arkansas Hills paleovalleys had >450 m of late Eocene relief. Trimble (1980) appealed to surface uplift at that time to explain conglomerate deposition. Eaton (2008) argued that today's SRM ranges are erosional remnants of their Laramide predecessors that sit atop the crest of an ~1500-m-tall epeirogenic dome (Fig. 11). Late Eocene rise of this dome, with the Sawatch anticline piggybacked atop, would increase local relief, thus raising transport energy, facilitating the cutting of paleovalleys, and depositing conglomerate long after the Laramide orogeny ended and before ignimbrite flare-up volcanism.

Laramide sediments in both the Denver and South Park–High Park basins consist of basin-spanning units that record contemporaneous basin subsidence; accommodation space was continually produced during the orogeny. When deposition resumed after the early-middle Eocene uncon-formities, accommodation space was no longer being created.

The late Eocene units in High Park illustrate this cut-and-fill architecture. They fill abundant paleochannel remnants that reveal a south- and east-flowing drainage system like the one that drained the Laramide orogeny (Fig. 1; Epis and Chapin, 1975; Scott, 1975; Epis et al., 1976; Hills and Dickinson, 1982). This drainage pattern persisted through the late Eocene and early Oligocene, but individual channels frequently shifted, cutting previous deposits and filling with younger material. For example, the Echo Park Alluvium is the oldest unit overlying the RMES (Fig. 2), but the Wall Mountain Tuff directly rests on the RMES in many places (Fig. 4B). The Tallahassee Creek Conglomerate overlies the Wall Mountain stratigraphically, but post–Wall Mountain erosion resulted in the Tallahassee Creek directly overlying the Echo Park in places and the RMES elsewhere (Fig. 2; Epis and Chapin, 1974; McIntosh and Chapin, 2004).

The sedimentary record of the Denver Basin tells a similar story (Fig. 2). In places, the Larkspur Conglomerate, an Echo Park equivalent, rests unconformably on the Dawson Arkose, with the Wall Mountain Tuff above. Elsewhere, pre–37 Ma erosion removed the Larkspur; the Wall Mountain directly overlies the Dawson (Fig. 2). Post–37 Ma erosion removed both the Larkspur and Wall Mountain units elsewhere, leaving the Castle Rock Conglomerate unconformably above the Dawson (Figs. 2, 4C, and 4D; Thorson, 2011; Keller and Morgan, 2016).

Gregory and Chase (1994) noted that Laramide crustal shortening should produce a Denver Basin deeper than 900 m—the thickness of the Denver Basin Group. They explained this seeming deficiency by ca. 54 Ma “emplacement of a buoyant subsurface load” under the basin (i.e., epeirogenic uplift). The Denver Basin Group is missing entirely from the northern basin; Cretaceous WIS units are overlain by the late Eocene White River Group (Tweto, 1979). Tweto (1975) interpreted this unconformity to reveal a Paleocene or Eocene rise of the Denver Basin. High Park is also missing the expected Laramide-age sediments, suggesting a similar history.

The relationships described here all suggest that central Colorado's east-southeast–flowing Laramide drainage system did not change appreciably during the Eocene, but a major change in drainage energy and style did occur. Active subsidence ceased in Laramide basins by 56–54 Ma, and erosion created unconformities. When deposition resumed at 41–37 Ma, the energy level had increased—before the growth of ignimbrite flare-up volcanic topography. These deposits exhibit a cut-and-fill architecture caused by the lack of accommodation space. What caused these changes plus contemporaneous Arkansas Hills exhumation? We favor Eocene epeirogeny, but we must entertain other possible causes before we can confidently draw that conclusion.

Molnar and England (1990) noted three possible causes for exhumation and conglomerate deposition: (1) drainage reorganization, (2) climate change, and (3) contemporaneous tectonism. Here we examine each candidate cause for Eocene central Colorado exhumation and conglomerate disposition.

Stream capture can increase a river's discharge, thereby increasing its erosional efficiency, leading to exhumation and/or conglomerate deposition. But, as noted above, central Colorado's drainage network (Fig. 1) changed little between the Laramide and the Oligocene (Scott, 1975; Epis et al., 1976); drainage reorganization is an unlikely cause of the area's Eocene changes.

Climate change can trigger accelerated erosion in a previously uplifted landscape. Workers have cited cooling events at the Eocene–Oligocene boundary (33.9 Ma), in the middle Miocene (ca. 15 Ma), and in the late Pliocene (ca. 3–4 Ma) as triggers for exhumation and conglomerate deposition (e.g., Molnar, 2004; Pelletier, 2009). But no abrupt change in climate that would likely trigger increased erosion coincides with central Colorado's Eocene changes (Fig. 12).

Three styles of tectonism could increase erosion in central Colorado: (1) crustal shortening, (2) crustal extension, and (3) epeirogenic surface uplift. As detailed above, Eocene crustal shortening did occur in Colorado but not in the Sawatch anticline; it cannot explain Arkansas Hills exhumation at that time. Similarly, high relief caused by extension fails as an explanation because the Rio Grande Rift formed later, in the Oligocene (Landman and Flowers, 2013; Ricketts et al., 2016).

Consideration of all possible mechanisms for Eocene central Colorado exhumation and conglomerate deposition leaves only one viable explanatory option—epeirogenic surface uplift.

Several previous low-temperature thermochronologic studies have been conducted in central Colorado: Why didn't they detect Eocene epeirogeny? Here we posit that one AFT study did detect epeirogeny in the Sawatch Range, and another is suggestive for the northern Front Range. Bryant and Naeser (1980) concluded that three Eocene AFT dates from the Sawatch Range (51 and 52 Ma on two Precambrian rocks east of Aspen and 46 Ma from the 66 Ma West Tennessee Stock; see Fig. 1) “may represent an uplift event of regional significance during the Eocene that hitherto has been unknown” (p. 163). Epeirogeny is the most likely cause of this Eocene cooling; the reason, once again, is that crustal shortening ended in the Sawatch during the Paleocene.

Kelley and Chapin (2004) documented many 54–44 Ma (Eocene) AFT dates in the northern Front Range, including 50 Ma and 48 Ma dates for two samples near the town of Lyons (Fig. 1). Cenozoic evolution of the Front Range is complex; Kelley and Chapin (2004) attributed Eocene cooling there to a combination of Laramide crustal shortening, heating by Colorado Mineral Belt plutons, and thermal insulation from thick, low-conductivity Cretaceous shales—not epeirogeny. But Hoblitt and Larson (1975) argued that Laramide deformation ended before 63 Ma near Lyons. Reheating by Colorado Mineral Belt plutons is possible, but the nearest exposed pluton is 12 km away (Cole and Braddock, 2009); therefore, reheating was likely modest. Eocene epeirogeny provides an equally, or more, plausible explanation for the Eocene AFT dates near Lyons.

Combination of these AFT dates, the interpretation that Florissant (see Fig. 1 for location) had reached its current elevation by 34 Ma (Meyer, 1992; Gregory and Chase, 1992), rapid, deep erosion of Sawatch range flare-up volcanoes (Lipman, 2021), the Denver Basin unconformity, and our Arkansas Hills data allow us to draw a perimeter around the part of central Colorado that has evidence of Eocene epeirogeny (the red dashed line on Fig. 1). It measures ~170 km west-east (from Aspen to Castle Rock) by ~120 km north-south (from Lyons and the West Tennessee stock to the Whitehorn pluton). The affected area is >2.0 × 10⁴ km², from the Sawatch Range to the Great Plains (Fig. 1).

Here we explore the many options that could explain how central Colorado could rise epeirogenically during the Eocene—both previously proposed mechanisms and a new hypothesis.

Previously proposed mechanisms for epeirogenic uplift (and their implied timing) include: (1) removal of subcontinental lithosphere by a flat Farallon slab (syn-to immediately post-Laramide, ca. 70–35 Ma; Bird, 1988; Spencer, 1996); (2) de-densification of the subcontinental lithospheric mantle and/or continental crust by slab-derived fluids followed by slab removal (syn-to immediately post-Laramide, ca. 70–35 Ma; Humphreys et al., 2003; Jones et al., 2015); (3) delamination of underplated, eclogitized ignimbrite flare-up lower crust (post–35 Ma; Karlstrom et al., 2012; Lipman, 2021); (4) lithospheric thinning due to RGR extension (post–30 Ma; McMillan et al., 2002); (5) geologically recent mantle heating (ca. 10–5 Ma; Aslan et al., 2010; Karlstrom et al., 2012; Rosenberg et al., 2014); and (6) normal forces due to asthenospheric convection (i.e., dynamic topography); <10–5 Ma; Moucha et al., 2008; Becker et al., 2014; van Wijk et al., 2018).

Considering only the timing they imply, mechanisms 1 and 2 are viable explanations for Eocene epeirogeny in central Colorado. Mechanisms 3 through 6 all predict surface uplift after the Eocene. Therefore, none of them can explain the central Colorado epeirogeny, and we do not consider them further here; these mechanisms could have operated in other places at other times, though.

Bird (1988) proposed the shear traction hypothesis in part to explain a thicker SRM crust than could be produced by Laramide crustal shortening. Subsequent work (e.g., Sheehan et al., 1995) revealed that the SRM crust is no thicker than that beneath the adjacent Great Plains; the SRM are not supported by Airy isostasy as Bird (1988) envisioned. A further difficulty is that the model predicts complete removal of continental lithosphere, which several studies have shown remained beneath the region after the end of the Laramide orogeny (see Jones et al., 2011). So, although shear traction could explain the epeirogeny's timing, current understanding of the regional geology makes it an unlikely candidate.

Lee and Grand (1996) documented anomalously low seismic velocities beneath the SRM. They concluded that the mantle lithosphere is thin with hot, upwelled asthenosphere beneath the mountains. Seismology cannot detect the timing of upwelling. Others have suggested mantle heating is recent (Mechanism 5 above). Recent heating cannot explain our observations, but asthenospheric upwelling beginning ca. 54–46 Ma could (see the next section).

Humphreys et al. (2003) confirmed that lowvelocity mantle underlies the SRM, but they attributed it to partial melt, not upwelled asthenosphere. They inferred that 200-km-thick lithosphere underlies the SRM; they explained high topography despite this thick lithosphere via hydration of the continental lithospheric mantle by fluids derived from the Farallon flat slab. Metasomatism de-densified the mantle lithosphere, causing epeirogenic uplift during the Laramide orogeny. A second epeirogenic episode followed at the end of the Laramide, caused by replacement of the sinking Farallon slab by asthenosphere—this timing fits our observations.

Jones et al. (2015) proposed a modified idea—that hydration de-densified the lower crust, not the mantle. This is an attractive explanation for the high elevation of the western Great Plains, which largely lack Cenozoic volcanism and are underlain by higher-velocity mantle.

The hydration hypothesis is appealing, but, assuming the Farallon slab dewaters uniformly, it predicts contemporaneous surface uplift every-where above the flat slab. If such uplift triggered Eocene exhumation in central Colorado, one expects rocks elsewhere to record similar exhumation. That prediction is not supported by thermochronologic data from Colorado's Cenozoic plutonic rocks. As discussed below, kilometer-scale exhumation occurred in the Oligo-Miocene in the Spanish Peaks area and in the Miocene for the “Gothic Dome” (Fig. 1). No previous hypothesis, not even the hydration hypothesis, can explain well this diachronous exhumation across the SRM, leading us to suggest a new alternative—serial mantle drips.

Houseman and Molnar (1997) considered the convective removal of mantle lithosphere via a Rayleigh-Taylor instability (i.e., a “mantle drip”) to be an almost “inevitable” consequence of orogenic crustal thickening. Their numerical models revealed that non-Newtonian viscosity in olivine could cause delays of tens of millions of years between the end of shortening and the onset of a mantle drip. Mantle drips have been invoked to explain epeirogenic uplift episodes around the world, from the Sierra Nevada to the Andes to Tibet (e.g., Kay and Kay, 1993; Houseman and Molnar, 1997; Farmer et al., 2002; Jones et al., 2004; Ducea et al., 2013).

Our drip hypothesis posits that lithospheric thickening accompanied Laramide crustal thickening beneath the Sawatch and Front Ranges; lateral flow of viscous mantle lithosphere toward this small perturbation caused it to grow, first slowly, then exponentially (Figs. 13A and 13B). When the negatively buoyant drip detached in the Eocene, replacement with less dense asthenosphere produced epeirogenic surface uplift in central Colorado (Fig. 13C), resulting in Arkansas Hills exhumation, unconformities in the South Park–High Park and Denver basins, and conglomerate deposition. Because subsequent cooling of the asthenosphere to form thermal lithosphere occurs over tens of millions of years, the surface height produced by epeirogeny persists to the present.

Lower lithospheric heterogeneity created by the sinking of one drip can initiate formation of an adjacent drip, as Garzione et al. (2014) proposed to explain progressive surface uplift of the Bolivian Altiplano from south to north during the Miocene. Eocene removal of a mantle drip beneath central Colorado could plausibly trigger subsequent drips around its periphery. Lord et al. (2016) invoked an Oligocene drip to explain formation of multiple small-volume, ca. 25 Ma, alkaline magmatic centers in the Spanish Peaks area, ~170 km southeast of our study area (Fig. 1). Landman (2016) and Kainz et al. (2021) documented kilometer-scale exhumation of the Spanish Peaks and their surroundings a few million years after pluton emplacement. Activity of an Oligo-Miocene drip could explain both observations (Fig. 13D).

The “Gothic Dome,” in the Elk and West Elk ranges near Aspen, abuts the region of Eocene exhumation (Fig. 1). AHe, ZHe, and AFT data (Garcia, 2011; Abbott et al., 2021) record >4 km of Miocene exhumation in the center of the ~100-km-diameter dome, with the magnitude of exhumation tapering to near zero outside the dome's perimeter. Samples surrounding the dome all have much older AHe dates. We suggest that an Eocene drip east of Aspen could have spawned a Miocene drip south-west of Aspen—the latter produced the Gothic Dome (Fig. 13D).

Does the area's lithospheric structure support the mantle drip hypothesis? Xenoliths derived from the shallow mantle lithosphere (<75 km depth) exist in 8 Ma basalt from Herring Park, in our study area (Fig. 3; Bailley, 2010). This observation does not preclude the drip hypothesis, as Houseman and Molnar's (1997) numeric models show that drips typically remove only 50%–75% of the mantle lithosphere. The seismic velocity structure of the area is complex (e.g., Schmandt and Humphreys, 2010). Workers variously conclude that the lithosphere-asthenosphere transition lies at 200 km (Humphreys et al., 2003), 150 km (e.g., Hansen et al., 2013), or <100 km (e.g., Lee and Grand, 1996; Levander and Miller, 2012); seismic knowledge of lithospheric structure beneath Colorado is not sufficiently mature to rule the drip hypothesis in or out. We suggest, though, that the activity of serial drips might explain the area's seismic complexity and the presence of a “lithosphere-asthenosphere mixing zone” between ~80–200 km depth as hypothesized by Karlstrom et al. (2012).

Our analysis of the Arkansas Hills' Whitehorn pluton using ZHe, THe, AFT, and AHe thermochronometers allowed us to trace its thermal history between 67 Ma emplacement at 5–7 km depth and the surface by 37 Ma. The pluton experienced two episodes of rapid cooling. The first, between 67 and 60 Ma, was due to thermal relaxation combined with Laramide-induced exhumation. The pluton temperature was 60–120 °C (~1.3–3.7 km depth) at 60 Ma. Slow cooling ensued for the next 6–14 m.y. until a second episode of rapid cooling began between 54 and 46 Ma. Exhumation that brought the pluton to the surface by 37 Ma caused this Eocene cooling event.

Laramide crustal shortening ended in the Arkansas Hills by 67 Ma; therefore, it could not have caused the Eocene exhumation episode. Two central Colorado Laramide basins, the Denver and South Park–High Park, stopped subsiding and developed unconformities at 56–54 Ma, precisely when exhumation began in the Arkansas Hills. When deposition resumed at 41–37 Ma, transport energy was higher than during the late Laramide, and accommodation space was limited, suggesting regional uplift during unconformity development. We attribute simultaneous Arkansas Hills exhumation and formation of unconformities to the onset of Eocene epeirogenic doming of central Colorado. The dome was at least 170 km east-west by 120 km north-south, affecting >2.0 × 10⁴ km² extending from the Sawatch Range to the western Great Plains.

Of the many hypotheses proposed to explain epeirogenic surface uplift in Colorado, hydration and de-densification of continental mantle lithosphere and/or crust by fluids released from the Farallon flat slab (Humphreys et al., 2003; Jones et al., 2015) is the best explanation for Eocene epeirogeny in central Colorado. But the hydration hypothesis has difficulty explaining a growing body of thermochronologic data that record kilometerscale exhumation in different parts of Colorado at different times. This diachronous exhumation is better explained by convective removal of mantle lithosphere (i.e., mantle drips) at different times in different places. We suggest that an Eocene drip triggered epeirogeny in central Colorado. Relief on the lithosphere-asthenosphere boundary created by that drip seeded an Oligo-Miocene drip in the Spanish Peaks area and a Miocene drip in the Elk and West Elk Mountains.

National Science Foundation (NSF) grant EAR-1559306 to Flowers and Metcalf provided support for new instrumentation during the course of this study that was used to acquire data in this project. Niazy thanks the NSF-funded RESESS program, run by UNAVCO, for the opportunity to work on this project. Abbott's ideas about central Colorado's evolution have benefitted from many discussions with Peter Molnar, Craig Jones, Lang Farmer, and Vince Matthews. We thank Peter Martin for help producing visualizations of HeFTy models that allow for visual tracking of individual time-temperature paths. We appreciate constructive reviews of an earlier version of this manuscript by Will Amidon and an anonymous reviewer and Associate Editor Greg Hoke's handling of the manuscript.