Volcanic rocks near Yampa, Colorado (USA), represent one of several small late Miocene to Quaternary alkaline volcanic fields along the northeast margin of the Colorado Plateau. Basanite, trachybasalt, and basalt collected from six sites within the Yampa volcanic field were investigated to assess correlations with late Cenozoic extension and Rio Grande rifting. In this paper we report major and trace element rock and mineral compositions and Ar, Sr, Nd, and Pb isotope data for these volcanic rocks. High-precision 40Ar/39Ar geochronology indicates westward migration of volcanism within the Yampa volcanic field between 6 and 4.5 Ma, and the Sr, Nd, and Pb isotope values are consistent with a primary source in the Proterozoic subcontinental lithospheric mantle. Relict olivine phenocrysts have Mg- and Ni-rich cores, whereas unmelted clinopyroxene cores are Na and Si enriched with finely banded Ca-, Mg-, Al-, and Ti-enriched rims, thus tracing their crystallization history from a lithospheric mantle source region to one in contact with melt prior to eruption. A regional synthesis of Neogene and younger volcanism within the Rio Grande rift corridor, from northern New Mexico to southern Wyoming, supports a systematic overall southwest migration of alkaline volcanism. We interpret this Neogene to Quaternary migration of volcanism toward the northeast margin of the Colorado Plateau to record passage of melt through subvertical zones within the lithosphere weakened by late Cenozoic extension. If the locus of Quaternary alkaline magmatism defines the current location of the Rio Grande rift, it includes the Leucite Hills, Wyoming. We suggest that alkaline volcanism in the incipient northern Rio Grande rift, north of Leadville, Colorado, represents melting of the subcontinental lithospheric mantle in response to transient infiltration of asthenospheric mantle into deep, subvertical zones of dilational crustal weakness developed during late Cenozoic extension that have been migrating toward, and subparallel to, the northeast margin of the Colorado Plateau since the middle Miocene. Quaternary volcanism within this northern Rio Grande rift corridor is evidence that the rift is continuing to evolve.

Latest middle Miocene and younger syntransform extension within the southern Basin and Range province (western USA, northwestern Mexico) is related to rapid migration of the Rivera triple junction past Baja Mexico to its current location between the tip of the Baja peninsula and mainland Mexico (Fig. 1; Dickinson, 2004). Evidence of this triple junction migration can be traced to late Cenozoic extension along the eastern margin of the Colorado Plateau, where the Neogene Rio Grande rift forms a prong of the Basin and Range (e.g., Prodehl and Lipman, 1989). Although the physiographic rift graben ends near Leadville, Colorado (Fig. 2), extensional deformation structures including high-angle block faulting and small, low-relief grabens filled with Miocene sedimentary rock of the Browns Park Formation continue west of the Gore and Park Ranges into Wyoming (Izett, 1975; Larson et al., 1975; Tweto, 1979). Locally, these extended areas are associated with late Cenozoic volcanic eruptions and volcanic fields of small to moderate size and varying degrees of alkaline character, including the Flat Tops Wilderness, Elkhead Mountains, and the Yampa volcanic field. These highly alkaline, mafic rocks have been broadly associated with Rio Grande rifting (e.g., Izett, 1975; Larson et al., 1975; Tweto, 1979), and are thought to represent low degrees of partial melting of previously enriched Proterozoic subcontinental mantle lithosphere (Leat et al., 1988, 1989, 1990; Thompson et al., 1990; Beard and Johnson, 1993).

The timing of alkaline volcanism in northwest Colorado related to late Cenozoic extension and Rio Grande rifting is approximately constrained by a few late Miocene to Quaternary K-Ar ages of compositionally diverse rock types (Izett, 1975; Larson et al., 1975; Thompson et al., 1993). Defining a high-precision chronology of this volcanism is essential to establishing any reliable spatial and temporal framework of late Cenozoic extension and volcanism. Because of their highly alkaline character, these rocks favor enrichment of incompatible elements such as the light rare earth elements (LREE) and study of their age and distribution could benefit future mineral exploration.

In this paper we report rock and mineral compositions and 40Ar/39Ar, Sr, Nd, and Pb isotope data for samples from the volcanic field in and around Yampa, Colorado. These results are the first Pb isotope, 40Ar/39Ar, and phenocryst compositional data reported from the Yampa volcanic field. Our high-precision 40Ar/39Ar geochronology yields Pliocene eruption ages and Pb isotopic data indicating rock derivation from a subcontinental lithospheric mantle that underwent silica-fluid metasomatism ca. 1.74 Ga. Using new and previously published data from northwest Colorado and southern Wyoming we demonstrate that the locus of alkalic volcanism during the past 15 m.y. has migrated toward the Colorado Plateau. We discuss implications of this migrating volcanism and its overall significance to late Cenozoic extension and the post-Pliocene evolution of the northern Rio Grande rift that includes the Leucite Hills, Wyoming.

The physiographic Rio Grande rift is marked by a series of discontinuous, sediment-filled grabens, half-grabens, and range-bounding normal faults consisting of three distinctive segments (Chapin, 1979; Ingersoll, 2001). The northernmost segment includes the geomorphic San Luis Valley, a large half-graben fault bounded by the towering (>4000 m elevation) Sangre de Cristo Mountains to the east (Fig. 2). North of the San Luis Valley, the Rio Grande rift is a smaller graben, bounded by the high-elevation Sawatch Range (>4000 m elevation) to the west and the lower elevation Mosquito Range to the east. Immediately north of Leadville is often considered the geographic terminus of the Rio Grande rift, and is where the last semicontinuous rift-like basin bounded by normal faults is exposed. However, block-faulting and downdropped grabens and half-grabens containing late Miocene sediment west of the Gore and Park Ranges and continuing north into Wyoming may represent less conspicuous Rio Grande rifting (Tweto, 1979). The areas between Leadville and the Leucite Hills are herein considered part of the incipient northern Rio Grande rift. Recent uplift of the Rocky Mountain epeirogen (e.g., Eaton, 2008) may obscure unequivocal physiographic expressions of a late Miocene rift-like depression north of Leadville, although an earlier episode of Cenozoic extension is clearly expressed in the North Park, Middle Park, and South Park basins (Fig. 2) that began developing shortly after Laramide arc-style deformation (Cole et al., 2010).

Much of the alkaline volcanism bordering the Colorado Plateau in northwest Colorado is associated with northwest-trending extensional fault blocks, and has long been attributed to Rio Grande rifting (Izett, 1975; Larson et al., 1975; Tweto, 1979; Leat et al., 1990, 1991; Beard and Johnson, 1993). The isotopic and chemical composition of their source rocks can provide insight into Rio Grande rifting and also constrain models of interacting asthenospheric and lithospheric mantle in proximity to the Colorado Plateau. For example, recent models show that late Cenozoic volcanic rocks along the southern margins of the Colorado Plateau become progressively younger and increasingly derived from the asthenospheric mantle approaching the plateau margin, and may reflect large-scale mantle dynamics that control the current size of the plateau, locus of magmatism, and development of relief (e.g., Wenrich et al., 1995; Roy et al., 2009; Crow et al., 2011; Karlstrom et al., 2012).

Both asthenospheric mantle (from Yellowstone) and subcontinental lithospheric mantle sources were proposed by Leat et al. (1991) for lavas from the Yampa volcanic field on the basis of elevated La/Ta ratios and a significant spread in 143Nd/144Nd ratios. Beard and Johnson (1993) followed this study by measuring Hf isotopes in lava samples from five separate areas within northwest Colorado sampled by Leat et al. (1988, 1989), including two samples from the Yampa volcanic field. The only samples from northwest Colorado that have 176Hf/177Hf isotope data consistent with a continental asthenospheric mantle source are the Yampa lavas, which have ΔɛHf values near zero, overlapping those of the ocean island basalt (OIB) field, and are similar to lavas of the Rio Grande rift near the Colorado–New Mexico border (Beard and Johnson, 1993). Otherwise, all volcanic rocks from northwest Colorado appear restricted to partial melting within the garnet peridotite field, from a subcontinental lithospheric mantle previously enriched by Proterozoic silica-fluid metasomatism within a shallower spinel peridotite field (e.g., Leat et al., 1991; Beard and Johnson, 1993).

Fundamental differences in lithospheric thickness beneath the southern and northern Rio Grande rift strongly influence the chemical and isotopic composition of late Cenozoic lavas (e.g., Lipman, 1969; Lipman and Mehnert, 1975; Prodehl and Lipman, 1989; Johnson and Thompson, 1991; Beard and Johnson, 1993; Johnson and Beard, 1993; McMillan et al., 2000), and temporal variations in volcanic rock composition provide important insight into the extent and dynamic evolution of the rift. On the basis of available K-Ar geochronology, Leat et al. (1989) noted three periods of alkaline volcanism in northwest Colorado, the earliest phase represented by Oligocene to early Miocene lamproitic lavas within the Middle Park basin (York et al., 1971; Larson et al., 1975; Thompson et al., 1993, 1997). A second volcanic episode was defined by the late Miocene and Pliocene minettes of the Elkhead Mountains, the Flatops Wilderness, and alkalic lavas near Yampa (Izett, 1975; Larson et al., 1975; Thompson et al., 1993). In the Yampa volcanic field, ages of 7.5–10 Ma were estimated from partially reset fission tracks in apatite separated from basement xenoliths (Izett, 1975) and from two samples with K-Ar ages of 5 and 6 Ma from basalt near Yampa (Thompson et al., 1993). A third volcanic period was noted on the basis of Quaternary mafic volcanic eruptions at Dotsero, McCoy Mountain, Triangle Peak, and Willow Peak (Giegengack, 1962; Larson et al., 1975; Leat et al., 1990). Lange et al. (2000) reported high-precision 40Ar/39Ar ages between 0.9 and 1.3 Ma for the highly alkalic rocks of the Leucite Hills (Carmichael, 1967), and it is the only volcanic field along the northeast margin of the Colorado Plateau characterized by high-precision 40Ar/39Ar geochronology. Establishing a similarly precise chronology of the eruptive sequences in northwest Colorado is necessary before rigorous testing of any models incorporating temporal constraints.


Samples of fresh, xenolith-free, volcanic rock were examined petrographically using polarizing and scanning electron microscopes, analyzed chemically by electron microprobe, and aliquots were prepared for geochemistry, 40Ar/39Ar geochronology, and Sr, Nd, and Pb isotope analysis. We sampled volcanic rock at six representative sites near Yampa, where numerous small volcanic necks, dikes, and lava flows occur as part of a small volcanic field west of the Park Range near Yampa (Fig. 3). Most of the volcanic rocks occur within or along the margins of small grabens and half-grabens containing sediment of the Miocene Browns Park Formation (Fig. 4). The lavas were forcibly emplaced, and most contain numerous shallow crustal xenoliths of the Browns Park Formation, ranging in diameter from a few millimeters to several meters, but also include centimeter-sized xenoliths of Precambrian basement rocks (Leat and Thompson, 1988) (Fig. 5).


The Yampa lavas have millimeter-sized phenocrysts of olivine and clinopyroxene and minor Ti-phlogopite and orthopyroxene in a vesiculated groundmass of plagioclase, clinopyroxene, iron oxides, and glass (Fig. 6). Olivine forms both tabular and skeletal crystals and the largest phenocrysts have been significantly replaced by chlorite. The larger clinopyroxene phenocrysts contain melt pools in their cores and have rims with numerous fine-scale, compositionally distinct bands, as indicated by petrographic observation (Fig. 6) and backscattered electron imaging (Fig. 7). The prominent sieve texture preserved in many of the larger olivine and clinopyroxene phenocrysts (Fig. 7) suggests that they are melting and that the fine-scale zoning of the rims is tracking chemical equilibrium with a melt phase during crystal growth. Vesicles observed in many samples contain secondary calcite.

Rock Compositions

Major and minor element concentrations were determined at the U.S. Geological Survey (USGS) in Denver on rock powders by quantitative X-ray fluorescence spectrometry, and trace element concentrations (including REEs) were determined by inductively coupled plasma mass spectrometry. For the trace element analyses, the whole-rock powders were digested overnight on a hot plate; external precisions of the reported trace element concentrations are ∼10% based on replicate analyses of the BCR-1 USGS standard.

Sr, Nd, and Pb Isotopes

The analytical techniques used for simultaneous, single-dissolution of U-Th-Pb, Rb-Sr, and Sm-Nd analysis on whole rocks for this study are similar to those reported in more detail by Tatsumoto and Unruh (1976), and Premo and Loucks (2000). Whole-rock samples were dissolved in 7 mL Teflon PFA vials with ultrapure concentrated HF + HNO3 and then spiked with a dilute mixed tracer of 205Pb-233U-236U-230Th as well as dilute mixed tracers of 84Sr-87Rb and 150Nd-149Sm. Samples were reheated to achieve isotopic equilibration. Lead was extracted using AG 1-X8 anion exchange resin in Teflon microcolumns using a very dilute HBr medium. Lead residues were dried in H3PO4 and loaded onto single Re filaments. The Pb laboratory contamination (blank) varied between 50 and 110 pg total Pb (average = 67 ± 8 pg), and had a measured composition of 206Pb/204Pb = 18.681 ± 0.064, 207Pb/204Pb = 15.432 ± 0.033, and 208Pb/204Pb = 37.720 ± 0.120 from multiple determinations. Uranium and Th were then extracted from the Pb eluent using AG 1-X8 anion exchange resin in a different, slightly larger, Teflon microcolumn using a 7N HNO3 medium, and residues were loaded onto Re filaments using dilute HNO3. Uranium blank levels were 15–25 pg, and Th blank levels were 1–6 pg. The effluent was then passed through a large (30 mL resin volume) column with AG 50W-X8 cation exchange resin, separating Rb, Sr, and the REEs. Strontium and Rb were further purified using a smaller Teflon column with AG 50W-X8 cation exchange resin; Sr was dried in H3PO4 acid and run on a single oxidized Ta filament, and Rb was run in a double rhenium filament configuration. Laboratory contamination levels of total Sr typically range between 30 and 300 pg. Rubidium blanks were typically five times smaller than those for Sr. Samarium was separated from Nd using AG 50W-X8 cation exchange resin and the α-isobutyric method of Lugmair et al. (1975), cleaned further using smaller Teflon columns with AG 50W-X8 cation exchange resin, then loaded with very dilute H3PO4 acid onto triple Ta filaments. Laboratory contamination levels of total Nd were between 30 and 250 pg; samarium blanks were typically 3–5 times smaller than those for Nd.

The Pb isotopes were measured using a Triton multicollector thermal ionization mass spectrometer (TIMS) in static mode using Faraday cups. The Pb isotope ratios were corrected for mass discrimination of 0.0010 ± 0.0003 per mass unit using data for National Institute of Standards and Technology (NIST) standards SRM-981 and SRM-982 measured at the same run conditions. The external uncertainty of the standard analyses was mainly due to mass fractionation effects, and was 0.08%, 0.12%, and 0.16% for 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb, respectively. The concentrations and isotopic ratios of U, Th, Rb, Sr, Sm, and Nd were determined on an automated, multisample, single-collector, VG Isomass 54R TIMS using the ANALYST programming of Ludwig (1992). Typical runs per element varied with desired level of precision, but for both Sr and Nd, a minimum of 240 measurements were collected to achieve a 0.003% error. Concentration uncertainties vary between ∼0.1% and ∼1.0%. All isotopic ratios were corrected for blank and instrumental mass fractionation, 87Sr/86Sr data were normalized to 86Sr/88Sr = 0.1194, and monitored for instrumental bias using the NIST SRM 987 standard; the mean value of 87Sr/86Sr for 30 analyses of the Sr standard during the course of this study was 0.710265 ± 10 (2σ). The 143Nd/144Nd data were normalized to 146Nd/144Nd = 0.7219 and monitored for instrumental bias using the La Jolla Nd standard, which yielded a mean value of 143Nd/144Nd = 0.511865 ± 10 (2σ) for 30 replicate analyses. The initial 87Sr/86Sr and 143Nd/144Nd values were calculated using measured Rb/Sr, Sm/Nd, and 40Ar/39Ar ages for the same samples. The ɛNd(0) values were calculated using present-day (143Nd/144Nd)CHUR = 0.512636, and (147Sm/144Nd)CHUR = 0.1967 (CHUR is chondritic uniform reservoir).

40Ar/39Ar Geochronology

The 40Ar/39Ar analyses were performed at the USGS in Denver, Colorado. Samples were prepared by crushing and isolating rock fragments of ∼1 mm3 from fresh rock free of obvious alteration and xenocrysts. The rock fragments were washed in deionized water and together with standards, were irradiated for 20 MW hours in the central thimble position of the USGS TRIGA reactor. Laser fusion of >10 individual Fish Canyon Tuff sanidine crystals (28.20 ± 0.09 Ma; Kuiper et al., 2008) at each closely monitored position within the irradiation package resulted in neutron flux ratios reproducible to ±0.25% (2σ). Isotopic production ratios and interfering nucleogenic reactions were determined from irradiated CaF2 and KCl salts and zero age K-silicate glass, and for this study the following values were measured: (36Ar/37Ar)Ca = (2.77 ± 0.03) × 10−4; (39Ar/37Ar)Ca = (6.54 ± 0.33) × 10−4; and (38Ar/39Ar)K = (1.29 ± 0.03) × 10−2. Cadmium shielding during irradiation prevented any measurable nucleogenic (40Ar/39Ar)K. The irradiated basalt samples and standards were loaded into numbered positions of a stainless steel planchette, placed into a laser sample chamber with an externally pumped ZnSe window, and evacuated to ultrahigh vacuum conditions in a fully automated stainless steel extraction line designed and built at the USGS in Denver. Using a 25W CO2 laser equipped with a beam homogenizing lens, the samples were incrementally heated and the liberated gas was expanded and purified by exposure to a cryogenic trap maintained at −140 °C and two hot SAES GP50 getters. Following purification the gas was expanded online into a Mass Analyzer Products 215–50 mass spectrometer in static mode and Ar isotopes were measured by peak jumping using an electron multiplier in analog mode. Data were acquired during 10 measurement cycles and time zero intercepts were determined by best-fit linear and/or polynomial regressions to the data. Data were corrected for mass discrimination, blanks, radioactive decay, and interfering nucleogenic reactions.

Rock Compositions

The Yampa volcanic rocks collected for this study are mafic (SiO2 = 41.5–47.5 wt%; MgO = 5–10 wt%), and alkalic (Na2O + K2O = 4–7 wt%), with high TiO2 (1.8–2.2 wt%) and P2O5 (0.8–1.0 wt%) (Table 1). These rocks plot as basanites, trachybasalts, and basalt with normative nepheline contents to 20% (Fig. 8) and have compositions similar to those reported in Leat et al. (1991). The Yampa lavas show slight positive correlations of Al2O3 with SiO2, whereas MgO and CaO are inversely correlated (Fig. 9A), indicating variable modal phenocryst abundances of olivine, clinopyroxene, and plagioclase relative to groundmass. Plots of TiO2 versus the other oxides (Fig. 9B), useful in evaluating geochemistry of high-Ti alkaline lavas (e.g., Carmichael et al., 1996), show that Ti is positively correlated with Nb (Fig. 10), consistent with a rutile-bearing source region (e.g., Foley et al., 2000). Elevated P2O5 contents support a P-rich source, possibly including apatite, and the metastable Ti-phlogopite preserved in sample 10YV-04A is a possible mantle source for the high Ti, Ba, and Rb concentrations (e.g., Schmidt et al., 1999) in these rocks.

The trace element composition of the Yampa samples have moderately elevated large ion lithophile element relative to OIB and highly enriched LREE concentrations relative to chondrites (Fig. 10). These new results are generally similar to those from early Miocene minettes from the Elkhead Mountains (Thompson et al., 1990) and the Quaternary alkaline basalts of northern Colorado (Leat et al., 1989), and slightly less enriched than the lamproites of the Leucite Hills (Mirnejad and Bell, 2006) (Fig. 10). Although the Yampa samples have elevated Rb/Nb, Ti/Nb, and La/Nb ratios relative to OIB, they plot on trajectories toward even higher ratios characteristic of some highly silicic minettes in the Elkhead Mountains (Fig. 10).

Mineral Compositions

Compositions of olivine, clinopyroxene, and phlogopite phenocrysts were determined from electron microprobe analyses, and both olivine and clinopyroxene phenocrysts exhibit significant compositional zoning (Table 2). Olivines have systematically Mg- and Ni-enriched cores and Fe- and Ca-enriched rims (Table 2; Fig. 11). Polarized light and backscattered electron imaging of compositional zoning in clinopyroxene phenocrysts (Figs. 6 and 7) is confirmed by the electron microprobe data, which show uniform enrichments in Si, Fe2+, and Na in cores indicating a small, but significant jadeite component (Table 2), and rims with marked enrichments of Ca, Mg, Al, and Ti relative to phenocryst cores. The clinopyroxene core and rim compositions require the following cation exchanges: Si ⇔ Ti, Fe ⇔ Mg, and NaSi ⇔ CaAl. Two quantitative microprobe traverses across one large (3 mm) phenocryst from sample 10YV-02 are representative of compositional variations observed among all analyzed samples (Fig. 12), with uniform compositions in the surviving clinopyroxene core and fine-scale, compositionally distinct bands along its rims.

Sr, Nd, and Pb Isotopes

The 206Pb/204Pb (17.5–18), and 208Pb/204Pb (37.3–37.6) ratios of the Yampa lavas (Table 3) are similar to Pb isotope data from numerous mafic, intermediate, and felsic basement samples of the Green Mountain arc terrane (Fig. 13A), a Proterozoic subduction zone or volcanic arc complex extending from northern Colorado to southwestern Wyoming (Hills and Houston, 1979; Premo and Loucks, 2000; Jones et al., 2011b). The high-grade metamorphic basement rocks of the Green Mountain arc terrane plot below the terrestrial Pb (bulk silicate Earth) evolution model of Stacey and Kramers (1975), consistent with time-integrated, hybrid mantle sources, and define a poorly constrained whole-rock Pb-Pb age of 1743 ± 64 Ma (mean square of weighted deviates, MSWD = 14), identical to U-Pb zircon ages from the regional basement rocks (Premo and Van Schmus, 1989). The Yampa lavas taken alone define a poorly constrained Pb-Pb age of 1998 ± 210 Ma (Fig. 13B), with Pb isotope values shared by mafic lower crust in this area, consistent with some lower crustal assimilation. Given the abundance of crustal xenoliths observed in the Yampa lavas, including sediments of the Miocene Browns Park Formation and high-grade gneisses, some degree of crustal assimilation seems unavoidable, yet the low 207Pb/204Pb (and 87Sr/86Sr) ratios suggest little upper crustal assimilation in these lavas.

The 143Nd/144Nd values of Yampa lavas range from 0.512327 to 0.512553 [ɛNd(0) values of 0 to −7] and initial 87Sr/86Sr values of 0.704–0.706 (Table 3), similar to those previously identified from other Yampa lavas (Leat et al., 1988). The Yampa lavas have a cumulative difference in radiogenic Nd of ∼7 ɛNd units, but no clearly defined trend in 87Sr/86Sr (Fig. 14A). This range in 143Nd/144Nd and 87Sr/86Sr ratios is consistent with melting some component of mafic basement of the Green Mountain arc terrane (Fig. 14B), and the range of ɛNd and ΔHf values also suggests a possible component of asthenospheric mantle (e.g., Leat et al., 1988; Beard and Johnson, 1993). A similar range in ɛNd values and restricted 87Sr/86Sr ratios is observed in the Elkhead Mountains, Yarmony Mountain, and younger Quaternary basalts from northern Colorado (Fig. 14B), but compared to the Yampa lavas their more negative ɛNd and ΔHf values indicate source regions restricted to the subcontinental lithospheric mantle (Beard and Johnson, 1993).

40Ar/39Ar Geochronology

Results of the Yampa whole-rock 40Ar/39Ar incremental CO2-laser heating experiments are presented in Table 4, and 40Ar/39Ar age spectra and 39Ar/40Ar versus 36Ar/40Ar isochrons are plotted in Figure 15; samples 10YV-01A and 10YV-01B were taken from different parts of one outcrop, and duplicate analyses of samples 10YV-02 and 10YV-05 are also shown. The 39Ar/40Ar versus 36Ar/40Ar isochrons for the Yampa samples (Fig. 15A) indicate trapped argon with 40Ar/36Ar values (ordinate intercept) significantly higher than the present atmospheric value of 298.56 (Lee et al., 2006). Excepting sample 10YV-06 (with insufficient isotopic differences between incremental heating steps to calculate a statistically valid isochron), samples plot along isochrons (MSWD < 2) with ages between 6.0 ± 0.3 Ma (10YV-01B) and 4.47 ± 0.11 Ma (10YV-05B) and have trapped 40Ar/36Ar ratios >300. The 40Ar/39Ar age spectra calculated assuming atmospheric 40Ar/36Ar values (298.56) for trapped argon are irregular (Fig. 15B) and lack 40Ar/39Ar plateau ages (3 or more consecutive heating steps with statistically identical [2σ] ages and combined total more than 50% of the cumulative 39Ar released). Only one sample (10YV-04B) has trapped 40Ar/36Ar ratios within error of present-day atmosphere, and has identical errorchron (5.04 ± 0.11 Ma) and near-plateau (5.06 ± 0.5 Ma) ages (Figs. 15A, 15B). Replotting all of the 40Ar/39Ar age spectra with the isochron-defined 40Ar/36Ar ratios (Fig. 15C) results in age spectra with 40Ar/39Ar age plateaus for nearly all samples with ages ranging between 6.08 ± 0.06 Ma (10YV-02A) and 4.59 ± 0.05 Ma (10YV-05A). Because the Yampa lavas have isochrons that indicate mixing with some extraneous argon with nonatmospheric 40Ar/36Ar values, the isochrons or age spectra using the isochron-derived trapped argon values most accurately define eruption ages within the Yampa volcanic field. The source of the extraneous argon may be related to assimilation of xenoliths of crustal rock prior to or during eruption. Sample 10YV06, with an integrated age of 22.5 ± 0.6 Ma, is the only true basalt sampled and contains half the magnesium of the other samples and is the only sample in which relict orthopyroxene was identified. Further work is required to determine the geochronological and geological significance of this sample.

Phenocryst Zoning and the Transition from Lithospheric Mantle to the Surface

Despite numerous crustal xenoliths contained in the Yampa volcanic rocks, no ultramafic xenoliths have been reported that could be useful in defining the pressure and temperature conditions of the subcontinental lithospheric source region. However, core-to-rim mineral zoning and preservation of phenocrysts of Ti-phlogopite provide indirect evidence on the nature of the source region. Phlogopite is a mineral that likely contributes to the petrogenesis of alkaline magmas in continental settings (e.g., Lloyd and Bailey, 1975; Edgar, 1987; Foley, 1992; Pilet et al., 2011) and is stable in lherzolitic mantle to depths of 180–210 km at temperatures between 800 and 1200 °C (e.g., Konzett and Ulmer, 1999). A source for the Yampa lavas within the phlogopite stability field is consistent with partial melting of garnet peridotite (e.g., Beard and Johnson, 1993) and further supported by seismic tomography, which indicates a lithospheric thickness of ∼150 km along the northeast margin of the Colorado Plateau (e.g., Dueker et al., 2001; Zurek and Dueker, 2005).

The core-to-rim zoning of both olivine and clinopyroxene phenocrysts in the Yampa lavas records distinct stages of mineral crystallization, one within the subcontinental lithospheric mantle and a second in contact with a compositionally varying melt during late Miocene extension. Relative to the phenocryst rims, the uniform compositions in phenocryst cores of clinopyroxene (enriched in Na and Si) and olivine (enriched in Mg and Ni) support residence at greater pressure. Olivine phenocryst rims have less Mg and Ni (Fig. 11), and clinopyroxene phenocryst rims have less Na and Al and are significantly enriched in Ti (Fig. 12). The larger phenocrysts, with a small jadeite component, are clearly melting, as evidenced by the numerous melt pools in most large grains. The fine-scale compositional bands developed along the clinopyroxene rims reflects changes in the melt compositions with which they were in contact over time scales of mineral growth. The compositionally uniform phenocryst cores, together with regular compositional trends preserved in the fine bands surrounding most phenocrysts, represent the overall transition from the subcontinental lithospheric mantle to the continental crust prior to eruption.

Late Cenozoic Migration of Alkaline Volcanism in Northwest Colorado

The nepheline-normative alkaline lavas of the Yampa volcanic field are volumetrically minor, but they are geodynamically important because they have geochemical signatures characteristic of incipient continental rifts (Carmichael et al., 1996). The alkaline and mafic lavas of the Yampa volcanic field were erupted between 4.5 and 6 Ma, and even at the scale of the field, the resolution of the 40Ar/39Ar data is sufficient to identify an overall southwestward temporal migration of magmatism (Fig. 3). At a slightly larger regional scale, and assuming that published K-Ar data are reliable, the Yampa volcanic field is one example of several occurrences of Pliocene to Quaternary alkalic volcanism that is progressively younger in the direction of the Colorado Plateau (Fig. 16). We interpret these rocks to represent partial melts from the subcontinental lithospheric mantle generated at depths >75 km in the garnet peridotite field. Because they have little isotopic evidence of crustal assimilation, preservation of these rocks forcibly emplaced at the surface almost certainly requires that they were following deep lithospheric pathways toward the surface. The northwest-trending high-angle Neogene faults associated with the Yampa and other Pliocene to Quaternary volcanic fields in northwest Colorado provide compelling evidence that volcanism is following and intruding into migrating zones of maximum strain developing within the continental lithosphere. These Pliocene and younger volcanic rocks, combined with the geologic and geochemical observations presented here, reflect incipient rifting of an otherwise stable subcontinental lithospheric mantle that is still active and consistent with global positioning measurements recording exceedingly slow yet measurable strain rates distributed broadly across this zone (Berglund et al., 2012). Quaternary rifting is limited to a zone subparallel to the northeastern margin of the Colorado Plateau extending at least as far north as the Leucite Hills (Fig. 16). The Quaternary volcanic rocks define the western limit of the Rio Grande rift, which diverges around the thick volcano-plutonic root of the southern Rocky Mountains volcanic field and tapers into thicker lithosphere at its north end.

Past and Present Magmatism within the Rio Grande Rift

Two regional episodes of late Cenozoic extension affected the Basin and Range province (e.g., Wernicke et al., 1987; Prodehl and Lipman, 1989; Dickinson, 2004), but only the younger episode seems to implicate the Rio Grande rift. An initial episode of Eocene to early Miocene extension is characterized by large calc-alkaline volcanic provinces and ignimbrite-forming calderas, including those of the southern Rocky Mountain volcanic field east of the Colorado Plateau (Lipman et al., 1971; Wernicke et al., 1987; Christiansen and Yeats, 1992; Dickinson, 2002; McIntosh and Chapin, 2004; Lipman and McIntosh, 2008). During that time, extension and igneous activity began and progressed from north to south in the northern Basin and Range province and began later and progressed from south to north in the southern Basin and Range province (Wernicke et al., 1987; Humphreys, 1995; Dickinson, 2002; Humphreys et al., 2003). Melting of mainly fertile Proterozoic (and minor Archean) lithosphere fluxed by devolatilization of subducted slab fragments may be responsible for volcanism of the western U.S. (e.g., Humphreys, 2009; Jones et al., 2011a). This volcanism generally coincides with Eocene extension and exhumation of metamorphic core complexes and Eocene to early Miocene calc-alkaline volcanic input into several large syntectonic basins in the northern Basin and Range province (Chamberlain et al., 2012).

A second episode of regional extension beginning ca. 20 Ma coincided with migration of triple junctions bounding the transform system (Fig. 1) that was subducting along the coast of western North America (e.g., Lipman et al., 1971; Ingersoll, 1982; Wernicke et al., 1987; Prodehl and Lipman, 1989; Jones et al., 1992; Atwater and Stock, 1998; Dickinson, 2002). As in other areas of the Basin and Range (e.g., Wernicke et al., 1987), this Miocene stage of extension in the Rio Grande rift corridor of Colorado and New Mexico is expressed by a transition from dominantly calc-alkaline magmatism to bimodal calc-alkaline and tholeiitic magmatism (e.g., Lipman, 1969; Lipman et al., 1971; Prodehl and Lipman, 1989; McMillan et al., 2000; Eaton, 2008; Chapin, 2012). In northwest Colorado the Elkhead Mountains, the Flat Tops Wilderness, and a few isolated eruptions apparently formed during this subsequent Miocene period of extension, and evidence of this extension was also recorded in the southern Colorado Plateau (e.g., Gonzales et al., 2010). Continued migration of the Rivera triple junction progressively affected the southern Basin and Range through the Pliocene and Quaternary, and is expressed by high-angle normal faults in northern Colorado and southern Wyoming, and coincided with an overall clockwise rotation in principal extension directions during the Pliocene from northeast-southwest to east-west (Prodehl and Lipman, 1989; Williams and Cole, 2007).

From 12 Ma to the present, bimodal volcanism within the Rio Grande rift corridor in New Mexico and southern Colorado reflects the ascent of asthenospheric mantle and its interaction with a thinning lithosphere (Lipman, 1969; Lipman and Mehnert, 1975). The first unequivocal asthenospheric magmas within the physiographic rift graben appeared ca. 10 Ma in the southern Rio Grande rift (McMillan et al., 2000) and ca. 6 Ma in the Rio Grande rift of southern Colorado (e.g., Lipman and Mehnert, 1975; Johnson and Thompson, 1991; Thompson et al., 2012). An asthenospheric source region for these basalts provides a simple distinction between rift-related magmatism and the earlier calc-alkaline magmatism and regional extension ∼10 m.y. earlier. The Rio Grande rift is therefore interpreted as a late Miocene and younger physiographic feature (e.g., Prodehl and Lipman, 1989), distinct from the Oligocene and early Miocene regional extension and volcanic activity imprinted over much of western North America. Although published 176Hf/177Hf isotope ratios support subcontinental lithospheric mantle sources for most lavas of northwest Colorado (Beard and Johnson, 1993), it seems likely that the heat necessary to melt the subcontinental lithosphere is derived from infiltrating asthenospheric mantle. With the possible exception of the Yampa lavas, magmas sourced in the asthenospheric mantle have not yet definitively appeared in the northernmost segment of the Rio Grande rift in Colorado and Wyoming, but this system is still evolving, as evidenced by the Quaternary alkaline lavas distributed along a zone northeast of the Colorado Plateau, including the Leucite Hills (Fig. 16). The lack of a definitive asthenospheric mantle Hf isotopic signal may be due to the dynamic nature of the entire melting process, as these highly alkaline lavas represent the first melt fractions from a previously veined lithospheric mantle that developed during an earlier period of lithospheric deformation and partial melting.

Lithospheric thickness is variable within the Rio Grande rift corridor but is notably thinner in the south than in the north (e.g., Prodehl and Lipman, 1989; Gao et al., 2004; Schmandt and Humphreys, 2010). Seismic tomography in the Rio Grande rift of southern New Mexico indicates a lithospheric mantle thickness of 45 km (e.g., Gao et al., 2004) and is thin enough to allow partial melts from the asthenospheric mantle to erupt at the surface (e.g., Harry and Leeman, 1995). Passive rifting alone does not account for thinning this initially thick (>100 km) lithosphere to 45 km, but more than 50% of the thinning could have occurred through mechanical abrasion and assimilation by the asthenospheric mantle (e.g., Gao et al., 2004; Byerly and Lassiter, 2012). In the northernmost Rio Grande rift (north of Leadville), the lithosphere is thicker than 100 km and present-day seismic tomography reveals anomalously slow P-wave traveltimes at a depth of 90 km, roughly coincident with an area approximately beneath the northern Rio Grande rift and extending from the northeastern margin of the Colorado Plateau into southern Wyoming (Yuan and Dueker, 2005; Schmandt and Humphreys, 2010). The seismic tomographic resolution is coarse, but seismic velocities in this region are indicative of melt at these depths, and may reflect asthenospheric mantle intrusion into, and current removal of, the subcontinental lithospheric mantle (e.g., Karlstrom et al., 2012). Although the available radiogenic isotope data support primarily lithospheric mantle sources for lavas of northwest Colorado (e.g., Beard and Johnson, 1993), there is some Nd and Hf isotopic evidence of an asthenospheric mantle source in the Yampa lavas, and continuing thinning of the subcontinental lithospheric mantle below northwest Colorado should lead to additional magmas originating in the asthenospheric mantle.

The Pliocene and younger alkalic volcanism along the northeast margin of the Colorado Plateau defines an overall northwest strike of the Rio Grande rift west of the Park Range that includes the Leucite Hills. This incipient northern Rio Grande rift projects broadly toward the Snake River Plain hotspot track at Yellowstone, where asthenospheric mantle is present at much shallower depths (Christiansen, 2001). Although the Wyoming craton, a thick Archean lithospheric root (e.g., Humphreys and Dueker, 1994), is between Yellowstone and the northern Rio Grande rift, the southwestward migration of Pliocene and younger alkalic volcanism indicates that thinning of, and intrusion into, this lithospheric mantle by asthenospheric magmas should occur in the geologic future. Large-scale sutures, such as the southward-dipping Cheyenne belt separating Archean and Proterozoic lithosphere near the Wyoming-Colorado border (e.g., Karlstrom and Houston, 1984; Zurek and Dueker, 2005), are recognized seismically as imbrications that have remained intact since the Proterozoic (e.g., Hansen and Dueker, 2009). Given that asthenospheric magmas are observed along Proterozoic sutures of the southern margin of the Colorado Plateau (e.g., Karlstrom and Houston, 1984; Magnani et al., 2004; Crow et al., 2011), such melts will eventually intrude along the Cheyenne belt.

Magmas unequivocally derived by partial melting of asthenospheric mantle within the Rio Grande rift are younger than 10 Ma, cut across preexisting physical boundaries, and overprint evidence of earlier regional extension and associated calc-alkalic igneous activity (see also Ingersoll, 1982; Baldridge et al., 1991). The distribution of Quaternary basalts in the Rio Grande rift indicates that magmatism is continuing and migrating toward the eastern margin of the Colorado Plateau. In part because of its location along the eastern margin of the Colorado Plateau, the Rio Grande rift has been interpreted as an incipient spreading center that will ultimately lead to plate separation and subsequent creation of a Colorado Plateau microplate (Coney, 1987). The proximity and colinearity of the alkaline Yampa rocks and neighboring Pliocene to Quaternary lavas in northwestern Colorado and Wyoming with the block-faulted Rio Grande rift farther south supports a hypothesis that asthenospheric upwelling is following zones of lithospheric weakness developed during late Cenozoic extension and this system is continuing to evolve and propagate along the northeast Colorado Plateau margin in a northwestward direction toward Yellowstone.

We gratefully acknowledge U.S. Geological Survey (USGS) colleagues Celeste Mercer, Heather Lowers, and Renee Pillers for assistance with the electron microbeam measurements and Jim Budahn for the X-ray fluorescence analyses. We thank the USGS TRIGA reactor staff for their assistance with sample irradiations, and Jennifer Sliney for introducing Cosca to the region near Yampa, Colorado. Discussions with and comments by USGS colleagues Jonathan Caine, Jim Cole, Ed du Bray, and Ed DeWitt helped formulate some of the ideas presented here, and Janet Slate provided valuable proofreading. We also thank two anonymous journal reviewers for constructive comments. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.