Large-volume travertine deposits in the southeastern Colorado Plateau of New Mexico and Arizona, USA, occur along the Jemez lineament and Rio Grande rift. These groundwater discharge deposits reflect vent locations for mantle-derived CO2, which was conveyed by deeply sourced hydrothermal fluid input into springs. U-series dating of stratigraphic sections shows that major aggradation and large-volume (2.5 km3) deposition took place across the region episodically at 700–500 ka, 350–200 ka, and 100–40 ka. These pulses of travertine formation coincide with the occurrence of regional basaltic volcanism, which implies an association of travertine deposits with underlying low-velocity mantle that could supply the excess CO2. The calculation of landscape denudation rates based on basalt paleosurfaces shows that travertine platforms developed on local topographic highs that required artesian head and fault conduits. Episodic travertine accumulation that led to the formation of the observed travertine platforms represents conditions when fault conduits, high hydraulic head, and high CO2 flux within confined aquifer systems were all favorable for facilitating large-volume travertine formation, which was therefore controlled by tectonic activity and paleohydrology. By analogy to the active Springerville–St. Johns CO2 gas field, the large volumes and similar platform geometries of travertine occurrences in this study are interpreted to represent extinct CO2 gas reservoirs that were vents for degassing of mantle volatiles into the near-surface system.


The southeastern Colorado Plateau in New Mexico and Arizona (Barker et al., 1996; Embid, 2009) and the Colorado Plateau–Rocky Mountain region in general (Crossey et al., 2009; Karlstrom et al., 2013) host numerous well-preserved travertine deposits (Fig. 1). This paper focuses on a set of remarkable large-volume (0.2–0.9 km3) travertine deposits with surface areas of 10–40 km2 and thicknesses ranging from 5 m to >60 m. These deposits are of particular interest because they provide a stratigraphic record of degassing of significant amounts of carbon dioxide (CO2) during discharge of groundwater from carbonic springs.

The large-volume travertine deposits of this study occur in the same general regions as Cenozoic basaltic volcanism, suggesting an overpressuring of the CO2-groundwater system with magmatic gases (e.g., Ballentine and Sherwood Lollar, 2002; Gilfillan et al., 2008, 2009). The deposits are preferentially located along faults of the western edge of the Rio Grande rift and along the Jemez lineament, a northeast-trending zone of late Cenozoic volcanic fields that transects the southeastern Colorado Plateau (Fig. 1). Two of the large-volume travertine deposits in New Mexico, Mesa Aparejo and the Riley travertine (Fig. 2), are in the proximity of an active magmatic system, both located near the margins of the Socorro magma body system, a sill-like intrusion at ∼19 km depth (Rinehart and Sanford, 1981; Ake and Sanford, 1988; Balch et al., 1997) that causes active uplift in the area (Fialko and Simons, 2001; Pearse and Fialko, 2010; Reiter et al., 2010) and long-lived high heat flow (Reiter et al., 2010). The travertine deposit at Mesa Del Oro is associated with Cenozoic basaltic volcanism (Fig. 2). The travertine deposits at Springerville, Arizona, are associated with an active modern CO2 gas field and a late Cenozoic volcanic field (Figs. 1 and 3).

Travertines of the Colorado Plateau have been the focus of recent attention in terms of models for diffuse mantle degassing through continents and degradation of groundwater quality due to input of endogenic fluids (Newell et al., 2005; Crossey et al., 2009; Karlstrom et al., 2013). The travertine record has also been linked to CO2 degassing during seismic events (Uysal et al., 2009), and travertine has been used as a natural laboratory for studying CO2 sequestration and leakage (Shipton et al., 2005; Dockrill and Shipton, 2010; Kampman et al., 2012). Some conclude that travertines may provide a potential terrestrial record of paleoclimate and paleohydrologic cycles (Szabo, 1990), and more specifically, that episodes of travertine deposition may take place in response to climatic warming (Kampman et al., 2012), and may provide a tool to date important climate transitions (Faccenna et al., 2008).

The goal of this paper is to evaluate both the history and processes of formation of large-volume travertine deposits of the southeastern Colorado Plateau. U-series dating is employed to determine when they formed, and their spatial and temporal distributions are used to discuss several potential influences on travertine deposition, including (1) tectonic forcings such as faulting, basaltic magmatism, and mantle degassing along the Rio Grande rift and Jemez lineament, (2) rates and patterns of long-term landscape denudation in this region, and (3) possible temporal associations with regional paleoclimate and paleohydrology records.


The term travertine is used here for continental limestones formed from discharge in springs and along streams (Ford and Pedley, 1996; Pentecost, 2005). A key to the formation of these fresh-water carbonate deposits is the degassing of carbon dioxide (CO2) from groundwater that is supersaturated with respect to calcium carbonate according to the following reactions (Pentecost, 2005; Crossey et al., 2006, 2009): 
It is the external CO2 that makes groundwater aggressive enough to dissolve the limestone. In modern travertine-depositing springs of the region, a significant component of the external CO2 has been shown to be derived from deep geological sources (Crossey et al., 2009; Karlstrom et al., 2013; Williams et al., 2013). Of the other volatiles carried with the CO2, He isotopes provide a tracer of potential mantle contributions (Ballentine and Sherwood Lollar, 2002). In particular, 3He/4He ratios of > 0.1 RA (RA = 3He/4He ratio in air = 1.4 × 10−6) show that the active travertine-depositing springs near the large-volume travertine deposits contain unequivocal evidence for the presence of mantle-derived volatiles (Newell et al., 2005). The sources of the deeply derived CO2 are varied and the water chemistry of modern travertine-depositing springs shows complex mixing of meteoric recharge, groundwater with long residence times, and the deeply derived fluids (Minissale et al., 2002; Newell et al., 2005; Crossey et al., 2006, 2009; Embid, 2009; Williams et al., 2013).

The morphology of travertine deposits is controlled by topography and the tectonic setting of the springs (Hancock et al., 1999; Pentecost, 2005). Fault-related fissure ridges, analogous to magmatic dikes, form on tensional fissures where travertine precipitates from spring orifices along a central fracture (Hancock et al., 1999; Chafetz and Folk, 1984; Pentecost, 2005; Brogi and Capezzuoli, 2009), and sometimes a fissure ridge is associated with spring mounds, circular domes of travertine that surround a central spring orifice (Pentecost, 2005). Spring mounds and terraced mounds as described by Chafetz and Folk (1984) and Pentecost (2005) require a hydraulic head and water under artesian pressure (Pentecost, 2005; Linares et al., 2010). If they self-seal due to rapid deposition or a drop in hydraulic head, the spring will migrate to a lower level, which leads to coalescing spring mounds and complex deposits (Pentecost, 2005; Linares et al., 2010). Coalescing spring mounds also form when the travertine-precipitating springs are aligned along a fault line (Pentecost, 2005). The travertine morphologies of the study areas in this paper comprise a fissure ridge, terraced mounds, spring mounds, and coalescing spring mounds; the latter form extensive planar travertine surfaces raised above the ground, referred to as travertine platforms herein.


Travertine deposits in New Mexico and Arizona occur along two major tectonic features, the Rio Grande rift and the Jemez lineament. The north-south–trending Rio Grande rift extends over more than 1000 km from Colorado to Mexico (Fig. 1; e.g., Keller and Baldridge, 1999; Reiter and Chamberlin, 2011). Structurally, it consists of a series of asymmetric Miocene, en echelon half-grabens that widen and become more diffuse to the south, until they interfinger with the Basin and Range province in south-central New Mexico (Keller and Baldridge, 1999; Reiter and Chamberlin, 2011). Extensional faults related to the extension of the Rio Grande rift were active from ca. 35 Ma to present, with the highest magnitude of extension of ∼10%–30% in the Miocene (Russell and Snelson, 1994). Normal faulting locally overprinted and reactivated compressional faults that formed Rocky Mountain uplifts during the Laramide orogeny (75–43 Ma) and so rift-bounding faults often record complex polystage movements (Keller and Baldridge, 1999; Cather, 2004; Seager, 2004; Minor et al., 2013). Ongoing extension in the Rio Grande rift is taking place at relatively slow rates of ∼0.1 mm/yr (Berglund et al., 2012).

A distinct crustal feature in the central Rio Grande rift is the Socorro magma body (Fig. 1), a sill-like intrusion at ∼19 km depth (Rinehart and Sanford, 1981; Ake and Sanford, 1988; Balch et al., 1997) that causes active uplift of the overlying region (Fialko and Simons, 2001; Pearse and Fialko, 2010; Reiter et al., 2010). Heat flow data suggest that the Socorro magma body is only the most recent manifestation of a much longer lived magmatic plumbing system within the rift (Reiter et al., 2010). The Socorro magma body is associated with the Socorro seismic anomaly, which causes numerous small magnitude events per year. The microseismicity and associated magmatism in various parts of the Rio Grande rift are thought to provide a source for mantle-derived CO2 and 3He (Newell et al., 2005; Williams et al., 2013).

In northern New Mexico, the Rio Grande rift intersects the Jemez lineament, which cuts across the Colorado Plateau, Rio Grande rift, and southern Great Plains (Aldrich, 1986; Chamberlin, 2007). The Jemez lineament is defined as a prominent belt of late Cenozoic basaltic volcanic fields (Fig. 1; Dunbar, 2005), the distribution of which has been attributed to the presence of a long-lived, northeast-trending intercontinental tectonic and magmatic zone that may have initially formed as a Proterozoic accretionary boundary (Aldrich and Laughlin, 1984; Karlstrom and Humphreys, 1998; Magnani et al., 2004; Karlstrom et al., 2005). This zone is directly underlain by a low-velocity sublithospheric mantle domain (Schmandt and Humphreys, 2010) that likely provided a source for the basaltic magmatism and appears to be tectonically active due to ongoing mantle-driven uplift, as suggested by bowed fluvial terraces, tilted basalt paleosurfaces, and river channels with high normalized steepness indices along the lineament where it extends out into the Great Plains (Wisniewski and Pazzaglia, 2002; Nereson et al., 2013).

Travertine Facies

A detailed facies analysis of the travertine deposits in the study areas was discussed more fully elsewhere (Priewisch et al., 2013) but is briefly mentioned here. The two important travertine facies from the study areas in New Mexico are a step-pool facies and a paludal facies. The step-pool facies forms along fissure ridges, on terraced mounds and spring mounds, and along streams. Several travertine lithologies and textures form in step-pool environments depending on location within these pools; these include (1) peloidal travertine, (2) horizontally laminated travertine, (3) drapes, (4) microterracettes, and (5) travertine breccia. The paludal facies represents marsh and lake margin environments, with a fluctuating water table, ponds, vegetated areas, and varying dry and wet conditions (Glover and Robertson, 2003; Pentecost, 2005; Crossey et al., 2011). Travertine lithologies in these environments include (1) carbonate mud or silt, (2) peloidal travertine, (3) peloidal travertine with plant remains, (4) horizontally laminated travertine, and (5) travertine breccia. Embid (2009) used facies classifications based on Fouke et al. (2000), Chafetz and Guidry (2003), Chafetz and Folk (1984), and Fouke et al. (2003) for the large-volume travertine deposits at Springerville in which both of these broad facies associations may have overlapped to produce vents, ponds, waterfalls, proximal and distal slopes, channels, and vegetated marshes.

Morphology and Facies of the Large-Volume Travertine Deposits

The study areas in New Mexico are Mesa del Oro, Mesa Aparejo, and the Riley travertine (North Mesa and South Mesa). Mesa del Oro is a Cenozoic basalt-capped mesa located near the southeastern boundary of the Colorado Plateau and associated with the Jemez lineament (Fig. 1). Large-volume travertine deposits formed in an area between two lava flows (Fig. 2), and field relationships in the northwestern part of the travertine deposit show that travertine overlies basalt. Travertine accumulation at Mesa del Oro led to the formation of a northern and a southern travertine platform (Fig. 2). A fissure ridge is part of the northern travertine platform that formed along an approximately north-northwest–south-southeast–trending fault (Figs. 2 and 4A; Priewisch et al., 2013). The travertine-precipitating groundwater discharged from spring orifices along a central fracture (Chafetz and Folk, 1984; Hancock et al., 1999; Pentecost, 2005; Brogi and Capezzuoli, 2009) and flowed down the slopes of the fissure ridge; this led to the formation of terraced mounds and the step-pool facies (Priewisch et al., 2013). The paludal facies can be found at the western edge of the travertine platform, representing the flow of the travertine-precipitating groundwater away from the fissure ridge in streams and channels, and the eventual formation of a marsh in the distal part of the deposit (Priewisch et al., 2013).

The large-volume travertine deposit at Mesa Aparejo is located on the eastern side of the Lucero uplift along a reactivated reverse fault called the Comanche fault that, along with the Cenozoic Santa Fe normal fault, separates Precambrian and older Paleozoic carbonates from younger Cenozoic rocks (Figs. 2 and 4B; Barker et al., 1996). Travertine-precipitating springs discharged along the Comanche fault, leading to the formation of coalescing spring mounds. The northeastern part of the deposit is actively quarried by New Mexico Travertine, Inc. (Austin and Barker, 1990), and the quarried travertines have supplied facing stone for the University of New Mexico in Albuquerque and the New Mexico State Capitol in Santa Fe.

Other large-volume travertine deposits, the Riley travertine, are located south of Mesa Aparejo. The Riley travertine consists of two travertine platforms separated by the Rio Salado (Figs. 1 and 2). Riley North Mesa overlies Paleozoic, Mesozoic, and Cenozoic rocks, and Riley South Mesa overlies Cenozoic strata (Fig. 2; Barker, 1983). Both travertine platforms are situated directly above the west margin of the Socorro magma body system (Figs. 1 and 2). Riley North Mesa formed when travertine-precipitating groundwater discharged along faults, leading to the formation of spring mounds that eventually coalesced (Fig. 4C). The travertine step-pool facies can be found throughout most of the travertine platform; the paludal facies formed in the southern part of the deposit. Riley South Mesa is located to the southeast of Riley North Mesa (Figs. 2 and 4D) and this travertine platform consists entirely of the paludal facies that formed in a wetland or marsh area.

The study area in Arizona is Springerville, where large-volume travertine deposits are located along the southwestern edge of the Colorado Plateau and the southwestern part of the Jemez lineament (Fig. 1). They consist of travertine platforms and travertine mounds that overlie Mesozoic and Cenozoic sedimentary rocks. Travertine-precipitating groundwater discharged in springs along normal faults, in particular the Coyote Wash fault (Figs. 3, 4E, and 4F), leading to the formation of spring mounds, which in some areas coalesced and formed large travertine platforms (Figs. 3, 4E, and 4F; Embid, 2009). At the spring mounds, the step-pool facies developed while the paludal facies formed in more distal areas. The large-volume travertine deposits at Springerville are associated with an active CO2 gas field (Fig. 3; Embid, 2009).


Mapping and geographic information system (GIS) volume analysis of the travertine extent was conducted on U.S. Geological Survey 7.5′ topographic maps and RGIS (New Mexico Resource Geographic Information System Program) digital orthophotos (1 m) in the field and digitally with ArcMap GIS ©1995–2014 Esri (Environmental Systems Research Institute, Inc.), using additional information of geologic maps compiled by Jicha (1958), Cather and Read (2003), Chamberlin (2004), and the New Mexico Bureau of Geology and Mineral Resources (2003). In order to assess the volume of travertine for each deposit, ArcMap was used to digitize the travertine extent on the digital orthophotos. The area was multiplied by the average thickness of each deposit to determine the volume. Travertine thicknesses were obtained through mapping, from borehole reports provided by the New Mexico Bureau of Geology and Mineral Resources, and published data (Moore et al., 2005; Embid, 2009).

Measured sections (Figs. 5 and 6; Table DR11) from the travertine platforms in New Mexico and Arizona (Table DR1 in the Supplemental File [see footnote 1]) were sampled to try to assess the duration of accumulations as well as individual ages. The travertine samples were cut into slabs, and dense layers of micrite or sparry calcite (Fig. 5) were microdrilled to provide powder for strontium isotope and U-series analysis. Except for Springerville, where Embid (2009) collected and dated several samples, only top and bottom samples of the measured sections were dated in order to constrain the depositional interval.

Travertine samples were dated with the U-series method at the Radiogenic Isotope Laboratory at the University of New Mexico using the methods described in Asmerom et al. (2010). This is a reliable dating method for measuring geologic age back to ca. 500–600 ka (Edwards et al., 1987) due to the fact that the system 234U -230Th returns to secular equilibrium within analytical resolution after ∼6–8 half-lives of the daughter isotope, 230Th, which has a half-life of 75,700 yr (Cheng et al., 2000). All of the samples are plotted on a uranium evolution diagram in order to visualize samples that have robust ages, samples that underwent uranium loss (evidence for alteration), and samples that were beyond U-series range, but still within the age range (ca. 1.5–2 Ma) amenable for 234U model ages (Fig. DR1 and Table DR2 in the Supplemental File [see footnote 1]). For samples outside of U-series range, model ages were calculated (Table 1) by using a range of assumed δ234U values corresponding to values from dated samples that had robust U-series dates within the same location (Tables DR3 and DR4 in the Supplemental File [see footnote 1]). This assumes that spring chemistry was similar through time in each area. If the minimum δ234U values of the area generated model ages within U-series range, they were rejected and the area δ234U mean and maximum values represent our preferred age brackets for these samples (Tables DR3 and DR4 in the Supplemental File [see footnote 1]).

Strontium isotopes of eight travertine samples were analyzed to help understand mixing relationships of different groundwater components (Clark and Fritz, 1997; Crossey et al., 2006). For this analysis, 10–50 mg of powdered travertine sample was dissolved in 0.5 mL of 7 N HNO3 and spiked with 1 mL 84Sr spike. The sample solution was then fluxed for 1 h; 0.2 mL of Sr specific cation resin was placed into 2 mL columns and conditioned with 3 N HNO3. Sr was dropped with 0.75mL H2O, then Sr separates were dried down and redissolved in 1 mL of 3% HNO3 for analysis on the Thermo Neptune multicollector inductively coupled plasma–mass spectrometer.


Travertine platform geometries and travertine ages that are shown in Figure 4 are described and discussed in the following regarding their extent, thickness, association with faults and active springs in the area, and new U-series ages.

The travertine deposits at Mesa Del Oro cover an area of 27 km2 and have an estimated travertine volume of 0.7 km3 (Fig. 4A). The travertine is thickest at the fissure ridge, to 65 m based on drill cores provided by the New Mexico Bureau of Geology and Mineral Resources. The fissure ridge is part of a complex fault system mapped to the southeast of Mesa del Oro (Fig. 2; Jicha, 1958). Well-preserved travertine vents were not found, but they were probably located in the area where the travertine is thickest. There are no active springs associated with the travertine platforms; however, dissolution caves may attest to past spring locations and groundwater activity within the northern travertine platform (Forbes and Stephens, 1994). The nearest travertine-precipitating CO2- and 3He-rich spring (Eddleman Spring) is ∼6 km to the northwest of the northern travertine platform (Fig. 2; Williams et al., 2013). The travertine deposits at Mesa del Oro locally overlie Cenozoic basalts that erupted from volcanic vents located at the northern edge (Cerro del Oro) and in the western part of the basalt mesa (Fig. 2; Jicha, 1958; Baldridge et al., 1987). Basalt flows have K-Ar ages of 3.5 ± 0.1 Ma, 3.4 ± 0.1 Ma, 3.1 ± 0.5 Ma, and 0.8 ± 0.5 Ma (Baldridge et al., 1987). These basalts flowed to the northeast and southeast along an arcuate paleodrainage subparallel to the modern Arroyo Colorado between 3.4 and 0.8 Ma (Fig. 2).

U-series ages from Mesa del Oro are 566 ± 68.5 ka, 360 ± 13.5 ka, 337 ± 8.0 ka, and 253 ± 5.4 ka (Fig. 4A; Table 1). Calculated model ages for two samples are slightly older, 700–590 ka and 760–650 ka. These results imply that travertine formation at Mesa del Oro occurred in two intervals, from 360 to 250 ka and from ca. 760 to 560 ka. Samples T4 and T5 from the western edge of the deposit have similar ages (566 ka and 700–590 ka, respectively) that overlap within error (Fig. 4A; Table 1). Two samples, T1 and T2, from the northwestern edge of the travertine platform gave stratigraphically constrained ages of 337 ka (bottom) and 253 ka (top), suggesting travertine deposition over 84 k.y. (Figs. 4A and 6A; Table 1). Strontium isotope values (87Sr/86Sr) of three travertine samples at Mesa del Oro are relatively nonradiogenic (Table 1) and compatible with hydrologic models where the majority of the water volume originates from surface recharge in the Lucero uplift (Goff et al., 1983). The δ234Uinitial values are also consistent with this interpretation (Table 1).

At Mesa Aparejo, the travertine platform extends over an area of 13.1 km2 and has an estimated travertine volume of 0.2 km3 (Fig. 4B). Coalescing mounds indicate that spring vents were aligned along the north-south–trending Comanche fault, one of the major rift-bounding fault systems that is interpreted to have acted as a conduit for spring water. A significant fraction of the CO2 may have originated from the Socorro magma body system and moved up a basement-penetrating Laramide fault system (Ricketts et al., 2014). Modern CO2- and 3He-rich springs associated with modern travertine deposits, e.g., Salado Arroyo Spring, are located nearby in modern washes along the same fault system to the north and south of the travertine deposit (Fig. 2; Newell et al., 2005; Barker et al., 1996). Volcanic vents located to the north of Mesa Aparejo erupted 4.1 ± 0.1 Ma and 3.7 ± 0.4 Ma (Bachman and Mehnert, 1978; Baldridge et al., 1987) and produced basalt flows that followed paleodrainages to the south (Fig. 2).

Travertine samples from the quarries Temple Cream, Sheherazade, and Vista Grande, located in the northeastern part of Mesa Aparejo (Fig. 4B), were dated. At Temple Cream, sample T10 has a robust U-series age of 254 ± 2.7 ka while sample T9 is in secular equilibrium and older than 2–1.5 Ma (Figs. 4B and 6B; Table 1). Two samples at Sheherazade, T11 and T12, gave U-series ages of 663 ± 169 ka and 709 ± 140 ka that overlap within error and are considered to be usable, though imprecise, ages (Fig. 4B; Table 1). A robust U-series age of 435 ± 13.5 ka was obtained from sample T7 at Vista Grande while a model age was calculated for sample T8 at the same location (Fig. 4B; Table 1) that ranges from 690 to 560 ka. For both quarry locations the younger travertine samples are inferred to be infillings that formed when groundwater was injected into the existing section due to a high hydraulic head. Embid (2009) also described these types of infillings in a travertine mound at Springerville where secondary infillings occurred in a pulse late in the lifespan of the mound. Two travertine samples from Mesa Aparejo, T7 and T10, have strontium isotope values (87Sr/86Sr) of 0.717823 and 0.720272 (Table 1). These highly radiogenic values indicate circulation of waters through granitic Precambrian basement, compatible with hydrologic models for mixing of meteoric recharge with deep fluids along faults of the western Rio Grande rift (Goff et al., 1983; Williams et al., 2013) and supported by the δ234Uinitial values for these samples (Table 1).

Riley North Mesa covers an area of 37.4 km2 and has an estimated travertine volume of 0.5 km3 (Fig. 4C). The travertine platform lacks well-preserved travertine vents. Rift-margin faults and the basement-Paleozoic unconformity probably served as conduits for the CO2-rich waters and springs discharged where faults intersected progressively down-cutting drainages. These faults are now concealed by the travertine but have been mapped to the south of Riley North Mesa (Figs. 2 and 4C; Lewis and Baldridge, 1994). Based on borehole data provided by the New Mexico Bureau of Geology and Mineral Resources, the travertine is thickest (to 22 m) in the southwestern part of the travertine platform. A model age ranging from 620 to 510 ka for the dated sample was calculated by using the mean and maximum value of δ234U of successful U-series ages from Mesa del Oro (Table DR3 in the Supplemental File [see footnote 1]) assuming that the travertine-precipitating groundwater that formed the deposits came from the same aquifer.

Riley South Mesa extends over an area of 18.2 km2, and has an estimated travertine volume of 0.2 km3 (Fig. 4D). The travertine platform consists of carbonate-cemented sandstone and centimeter- to meter-scale travertine layers intercalated at the top with pedogenic carbonate. Barker (1983) described travertine thicknesses of typically 4–6 m and as much as 15 m. Riley South Mesa is slightly tilted to the west (Barker, 1983) and lacks well-preserved travertine vents. This section has complex age-height relations but deposition took place from 288 to 138 ka. Individual ages of dated samples are shown in Figures 4D and 6D and Table 1. A modern example of a CO2- and 3He-rich travertine-depositing spring is the Rio Salado Spring, located in the drainage between the two travertine platforms (Fig. 2; Newell et al., 2005), which discharges from a limestone aquifer of the Pennsylvanian Madera Group, called Madera Limestone hereafter (Rawling, 2005). Three samples from Riley South Mesa, T14, T15, and T16, have strontium isotope values (87Sr/86Sr) of 0.710328, 0.709553, and 0.709127, respectively (Table 1). These values suggest mixing of dominantly meteoric waters with some input of deeply circulated waters; this is supported by the δ234Uinitial values (Table 1).

The travertine deposits at Springerville occur over an area of 34 km2 and have an estimated travertine volume of 0.9 km3 (Figs. 4E, 4F; Embid, 2009). Travertine vents on the Salado Platform have pronounced shield geometries (Fig. 7A) and central orifices (Fig. 7B) and are located at the distal end of the Coyote Wash basalt flow, which was dated as between 2.94 ± 0.14 and 3.67 ± 0.12 Ma (Embid, 2009). The Springerville volcanic field was active from 8.97 ± 0.19 Ma until 0.31 ± 0.07 Ma (Condit and Connor, 1996) and volcanic activity overlapped with travertine deposition at 0.35 ± 0.01 Ma (Embid, 2009). Travertine deposition is still occurring around Salado Springs, where the spring water contains both CO2 and mantle-derived 3He (Fig. 3; Gilfillan et al., 2008; Embid, 2009). The travertine deposits at Springerville are located above and along the western boundary of a commercially important CO2 gas reservoir, the Springerville–St. Johns field (Figs. 3, 4E, and 4F; Moore et al., 2005; Embid, 2009).

At Springerville, Embid (2009) obtained travertine ages ranging from 51 to 354 ka (Fig. 4F; Table 1). The dated top and bottom samples of a measured section at the largest travertine platform are 51 ± 0.4 ka (T18) and 73 ± 0.6 ka (T17), respectively, and represent semicontinuous travertine deposition spanning 22 k.y. (Figs. 4F and 6E; Table 1; Embid, 2009). Two other samples (T19, T20) from vents on the same travertine platform have ages of 292 ± 12.1 ka and 354 ± 8.8 ka (Fig. 4F; Table 1).

Summary of Travertine Ages

In this study, mainly tops and bottoms of travertine sections were dated in order to constrain time intervals of travertine formation. Other researchers who dated travertine (e.g., Sturchio et al., 1994; Faccenna et al., 2008; Sierralta et al., 2010) obtained individual ages ranging throughout the Quaternary, while our study shows that large volumes of travertine in New Mexico and Arizona formed episodically at 700–500 ka, 350–200 ka, and 100–40 ka (Fig. 8; Table 1). One sample that is older than 1.5 Ma indicates travertine formation at an earlier interval (Table 1), and travertine deposition is still occurring around some spring vents. Nevertheless, the grouping of U-series dates is viewed as a record of episodes of large-volume travertine deposition in the region. As noted here, U-series ages show several general types of relationships relative to travertine stratigraphy. In some sections, e.g., at Mesa del Oro and Springerville, U-series ages are in agreement with stratigraphic position, and indicate that the vents or mounds were active for several tens of thousands of years, with deposition rates of ∼10 cm/k.y. (Fig. 6A) to 1 m/k.y. (Embid, 2009), respectively. In other cases, e.g., at Mesa Aparejo and Springerville, U-series infilling ages of travertine veins are younger than adjacent stratigraphic layers, indicating times when head was high enough to inject groundwater into fractures within existing travertine mounds or travertine platforms. This suggests that the same spring vents could be active at multiple times and that hydraulic head was high enough at a later time than during the initial mound accumulation to allow deposition at about the same elevation. Model ages are less precise in terms of durations and timing of episodes, but they demonstrate that many of the major travertine platforms were active in the interval of 700–500 ka.


This study attempts to evaluate possible drivers for episodic late Quaternary travertine deposition by examining potential tectonic drivers (faulting, magmatism, mantle degassing), as well as paleohydrologic drivers.

Tectonic influences on episodic travertine formation are documented at a local scale by the nearly ubiquitous association of travertine locations and major fault networks (Figs. 1–4). Movement along the Comanche fault allowed for travertine accumulation at Mesa Aparejo (Figs. 2 and 4B; Callender and Zilinski, 1976; Austin and Barker, 1990; Barker et al., 1996). A dominance of north-south subvertical calcite-filled extensional fractures in the travertine quarries at Mesa Aparejo that are compatible with the approximately east-west–trending extension of the rift were reported in Ricketts et al. (2014). These fractures were active from ca. 2 Ma to ca. 250 ka, suggesting upward flux of hydrothermal fluids over long periods of time despite potential clogging due to calcite precipitation (Curewitz and Karson, 1997; Ricketts et al., 2014). Travertine at Mesa del Oro precipitated from a fissure that is part of a complex extensional fault system to the southeast of the travertine platform (Figs. 2 and 4A; Jicha, 1958; Priewisch et al., 2013) and associated with the east-west extension of the Rio Grande rift. The travertine platform at Riley North conceals a number of faults (Figs. 2 and 4C) mapped by Lewis and Baldridge (1994) that might have acted as potential pathways for CO2-charged groundwater. At Springerville, the biggest travertine platform formed along the Coyote Wash fault (Figs. 3, 4E, and F4F; Embid, 2009).

At much longer temporal and spatial scales, the association of travertine depositing springs and travertine platforms with regions of low mantle velocity (Fig. 9) suggests that diffuse mantle CO2 degassing may be an important source for the external CO2 necessary to deposit large volumes of travertine (Newell et al., 2005; Crossey et al., 2009). Episodes of travertine formation are associated with regional volcanism (Fig. 8) and underlying low-velocity mantle (Fig. 9), suggesting that magmatic degassing of CO2 in addition to diffuse mantle degassing (e.g., Karlstrom et al., 2013) contribute to large-volume travertine deposition. This supports the interpretation that CO2 in many Colorado Plateau travertines and gas fields is in large part of magmatic origin and ultimately mantle derived. For example, the large-volume travertine deposits at Springerville are associated with the actively mined Springerville–St. Johns gas field that contains both CO2 and He. CO2 is often considered to be a carrier gas for He (Sherwood Lollar et al., 1997; Ballentine et al., 2001); this is reinforced by a mean 3He/4He value of 0.43 RA (Gilfillan et al., 2008; Embid, 2009). This value indicates that ∼20% of the He is mantle derived and suggests that a significant proportion of the CO2 also likely originates from the mantle. Similarly, other large-volume travertine deposits in New Mexico and Arizona are associated with nearby springs that also contain mantle-derived He and, by inference, mantle-derived CO2 (Figs. 2 and 3; Embid, 2009; Williams et al., 2013). The association of mantle-derived He and magmatically derived CO2 is also present in some CO2 gas fields in Colorado and Texas (Ballentine et al., 2001; Ballentine and Sherwood Lollar, 2002; Gilfillan et al., 2008, 2009).These fields as well as most of the travertine-depositing springs are low in methane (Newell et al., 2005; Gilfillan et al., 2008), such that upward transport of CO2 and He in travertine-depositing springs is often independent of petroleum migration and instead reflects upward movement of carbonic fluids through the crust and along faults (Ballentine et al., 2001).

Neither of these tectonic associations, faults and mantle degassing, completely explains the episodes of travertine formation at 700–500 ka, 350–200 ka, and 100–40 ka. However, these intervals overlap with times of Quaternary basaltic volcanism (Fig. 8; Dunbar, 2005; Embid, 2009). This is compatible with models showing that magmatism and melt flux through the lithosphere is accompanied by high regional CO2 flux (Fig. 10).

The landscape position of travertine deposition relative to local base level provides clues on the paleohydrology at the time of deposition. Travertine-depositing springs are expected to discharge into drainage bottoms except in confined systems, where artesian pressure may push waters above the local base level. Rates of landscape denudation from inverted topography were determined from dated basalt flows as well as travertine platforms that are now high in the landscape (Fig. 7C). In order to calculate denudation rates, topographic profiles created in ArcMap GIS were used to measure the difference in elevation from the base of the basalt flow or travertine platform to the nearest stream base level, which gives the denudation magnitude, and the denudation magnitude was then divided by the age of the basalt or travertine sample (Fig. 2; Table 2).

Denudation rates at Mesa del Oro show steady denudation of ∼70 m/m.y. over the past 8 m.y. with variation of denudation rates over the past 4.5 m.y. between 44 and 75 m/m.y. (Fig. 11; Table 2). The basalt flows just southeast of the Arroyo Colorado valley have the same arcuate shape as the modern arroyo and are inferred to mark the paleochannel of an ancestral Arroyo Colorado (Fig. 2). Similarly, the basalt flows of the Rio Puerco valley followed paleochannels draining into an ancestral Rio Puerco (Fig. 2) that was located more to the west than the modern Rio Puerco (Love and Connell, 2005). Denudation rates based on the height of dated basalt samples near Mesa Aparejo relative to the modern Rio Puerco range from 75 to 98 m/m.y. over the past 4.1 m.y. and are similar to rates at Mesa del Oro (Fig. 11; Table 2). In both locations, dated travertine samples (T2, T4, T7, T9, and T11) are much higher in the landscape than corresponding basalt paleosurfaces of comparable age and give unreasonably high apparent denudation rates, indicating that they formed on elevated travertine platforms well above local base level (Fig. 11; Table 2). Similar hydrologic settings are known in the modern landscape where active travertine is depositing 100–200 m above local base level on elevated travertine platforms because of high head in confined aquifer and fault systems (Cron, 2011; Crossey et al., 2011). These analogies imply that episodes of travertine deposition represent times of high artesian head in more than one location and likely across the region. This concept is also well displayed near Springerville, where Embid (2009) used travertines associated with gravel terraces to calculate incision rates for the Little Colorado River that averaged 40–50 m/m.y. from 7 Ma to 100 ka and increased to 320 m/m.y. over the past 100 k.y. (Fig. 12). Travertine formation well above the local river-defined base level occurred at 350 ka, 200 ka, and 100 ka (Embid, 2009), showing that the hydraulic head was high during those times, and overlaps with regional episodes of high head and travertine accumulation in New Mexico.

Paleoclimate and paleohydrology influences provide another possible control on episodic travertine formation. The travertines of this study represent groundwater discharge deposits from regional confined aquifers; hence they provide information on the timing and magnitude of changes in the hydrologic budget related to recharge and discharge, i.e., paleoclimate. Previously workers have proposed correlations between travertine accumulation and climate cycles, but different papers have different conclusions. Pigati et al. (2011) found that groundwater-fed wetland deposits including carbonates formed during wet glacial periods in the Mojave Desert (California, USA), due to fluctuating groundwater levels during the Pleistocene driven by synoptic-scale climate changes. Faccenna et al. (2008) dated travertines from central Italy and suggested that they may provide a tool to date important climate transitions over the past 140 k.y., but that fault activity also played a role in travertine formation. Kampman et al. (2012, p. 352) dated travertine mounds in Utah (USA), correlated the times of travertine formation to existing climate records, and stated that travertine formation occurred within 2000 yr of the glacial to interglacial transitions over the past 140 k.y., suggesting that CO2 leakage increased at those times as a result of fracture openings potentially caused by “changes in groundwater hydrology” with resulting increased leakage from intermittent CO2 gas caps. Contrasting views regarding paleoclimate conditions were expressed by Brogi et al. (2010), who reported travertine deposition mainly during dry glacial periods; they emphasized the importance of tectonic activity to facilitate travertine precipitation. De Filippis et al. (2013) compared travertine deposits at Tivoli (central Italy) and at Pamukkale (Turkey), and concluded that travertine formation occurred during warm and humid periods and was modulated by interactions between paleoclimate, fluid discharge, and fluid-driven tectonics. Özkul et al. (2013) studied travertine deposits in the Denizli Basin (Turkey) that formed both during warm and wet as well as dry and cold periods, and concluded that travertine precipitation is more a function of tectonic activity and not related to climate. Our study is not one of high enough resolution to resolve the issue. However, we note that the travertine record of this study shows some possible associations of wet (glacial) times as recorded by the marine oxygen isotope record (Fig. 8), and we also note that travertine deposition took place at times of high hydrologic head, implying wet times.

In our view, the best explanation for travertine episodicity involves interplay between tectonic controls on the CO2 flux and paleohydrologic controls on groundwater head. The large-volume travertine deposits in the study areas are interpreted mainly to represent a surface manifestation of mantle degassing focused through fault conduits. The travertines and associated springs with mantle volatiles (Newell et al., 2005; Williams et al., 2013) are located along the Rio Grande rift and Jemez lineament, both regions that are underlain by low-velocity mantle (less dense and more buoyant) (Fig. 9; Schmandt and Humphreys, 2010); they are related to basaltic volcanism, most notably at Mesa del Oro, Mesa Aparejo, and Springerville (Figs. 2 and 3), carrying mantle volatiles such as CO2, which is critical for travertine formation. Dynamic mantle processes have been linked to lithospheric extension, faulting, basaltic volcanism, and migration of mantle gases to the surface (Newell et al., 2005; McMillan et al., 2006; Karlstrom et al., 2008; Crossey et al., 2009; Jayko, 2009; Karlstrom et al., 2012). The Socorro magma body system, also located above low-velocity mantle, is the inferred CO2 source for the Riley travertine and the travertine deposits at Mesa Aparejo (Figs. 1 and 2). By analogy to the active Springerville–St. Johns gas field and its associated large-volume travertine deposits, it is inferred that all of the ancient large-volume travertine deposits in New Mexico are extinct CO2 gas fields (Fig. 9). A conceptual model of the tectonic setting in the Rio Grande rift, on the Colorado Plateau–Jemez lineament, and associated travertine deposition is shown in Figure 10: rift-related normal faults serve as conduits that convey deeply derived magmatic and/or hydrothermal fluids and CO2 to the surface where they emerge as travertine-depositing springs. Magmatic systems originating in the upper mantle lead to the formation of the Socorro magma body and volcanic activity, e.g., along the Jemez lineament.

Paleohydrology controls are considered to be equally important as tectonic controls, but they are not simply related to glacial-interglacial oscillations. Instead, evidence from this study focuses on high groundwater head during travertine formation, which implies that groundwater originating from a confined aquifer system had significant artesian head to ascend along faults and discharge at fault-controlled locations. Therefore, interpretations at individual sites need to examine groundwater flow paths and fault conduits. Regional aquifers in the study areas in New Mexico and Arizona are the Permian San Andres–Glorieta aquifer and Pennsylvanian Madera Limestone. Recharge areas are the Lucero uplift and the Colorado Plateau (Goff et al., 1983; Baldwin and Anderholm, 1992; Rauzi, 1999; Rawling, 2005), and it is inferred that the groundwater that formed the large-volume deposits in the study areas discharged from these aquifers. Hydrologic studies show groundwater mixing and complex flow paths at Mesa Aparejo and Mesa del Oro because recharge areas are in the Lucero uplift to the east and the Zuni Mountains to the west. Goff et al. (1983) found mixing of different types of groundwater along the Lucero uplift, in accordance with δ234Uinitial values and strontium isotope values of the samples from Mesa Aparejo, Mesa del Oro, and Riley South Mesa (Table 1). The radiogenic strontium isotope values at Mesa Aparejo and low δ234Uinitial values reflect input of old, deeply circulating groundwater that interacted with Precambrian basement, while nonradiogenic strontium isotope values and higher δ234Uinitial values at Mesa del Oro and Riley South Mesa represent input dominantly of shallow groundwater that reacted with Madera Limestone (Forbes and Stephens, 1994; Barker et al., 1996; Rawling, 2005; Crossey et al., 2006; Hogan et al., 2007; Burnside, 2010). The recharge area for the San Andres–Glorieta aquifer in Arizona is the Mogollon Rim, and groundwater flow is from the southwest to the northeast (Akers, 1964; Rauzi, 1999). Shallow and deeply circulating groundwater mixes in the study area at Springerville, rises along faults, and discharges as travertine-precipitating springs (Embid, 2009). More local thermally induced contributions to the groundwater system reflect local volcanic activity and hydrothermal circulation (Ingebritsen et al., 2008). Overall, hydraulic head is a response to (1) increased precipitation in the recharge area, (2) a rise of the water table within the aquifer, (3) a several thousand year response of distant springs to head change (Zhu et al., 1998; Sanford, 2002), and (4) local fault-related hydrothermal pressure (Ingebritsen et al., 2008; Uysal et al., 2009). There is clearly a connection between climate and recharge (Zhu et al., 1998; Sanford, 2002; Pigati et al., 2011), leading to the conclusion that large volumes of travertine formed when recharge was high. Thus, artesian springs and travertine episodicity are likely to be a more complex function of wet and dry intervals.


U-series ages show that the formation of large volumes of travertine in New Mexico and Arizona occurred episodically at 700–500 ka, 350–200 ka, and 100–40 ka. Episodes of travertine deposition require both high flux of CO2 and large amounts of groundwater discharge as well as the availability of faults, which serve as conduits for the CO2-charged groundwater toward the surface. These episodes of travertine formation overlap with episodes of magmatic activity, offering the permissive explanation that high CO2 flux takes place during basaltic magmatism and associated high seismicity such as exemplified by the modern Socorro magma body system. Denudation rates show steady lowering of the land surface over 8 m.y. in the study areas in New Mexico, varying from 44 to 98 m/m.y., while incision rates at Springerville average 40–50 m/m.y. over the past 7 m.y. with an increase over the past 100 k.y. Travertine platforms that stand high in the landscape today do not track base-level lowering; this indicates that the travertine-precipitating springs discharged during episodes of high hydraulic head even as erosion lowered the landscape. Thus, it can be inferred that episodes of high head resulted from an interplay of wet climate and high hydrothermal pressures, as filtered through complex artesian aquifer systems. We did not observe a simple correspondence of travertine accumulation episodes to glacial- and interglacial cycles in the data set of this study. The accumulated travertine volume of the study areas is 2.5 km3 (minimum) and by analogy to the active CO2 gas field at Springerville, it is inferred that other travertine platforms in New Mexico represent past degassing of CO2 reservoirs. Large-volume travertine deposits in New Mexico and Arizona provide important laboratories for continued studies of natural CO2 sequestration and leakage, as well as tangible mantle to surface system links.

The present work benefited from the input of Ryan Crow, who provided valuable assistance to the model age calculations. U-series dating was performed at the radiogenic isotope laboratory at the University of New Mexico. We thank the staff of the New Mexico Bureau of Geology and Mineral Resources for making travertine cores from Mesa del Oro, Riley North Mesa and Riley South Mesa available. We extend our thanks to Jim Lardner of New Mexico Travertine, Inc., for access to the travertine quarries at Mesa Aparejo, and to different parties at Mesa del Oro: Bob Worsley, owner of NZ Legacy LLC, for access to his property and travertine quarries; Jim Harrison, manager of NZ Legacy LLC, for valuable information about the travertine deposit; and rancher Mark Chavez for access to his land. We thank Andrea Brogi and an anonymous reviewer for their feedback and suggestions, which greatly improved the quality of the manuscript, and Drs. Tim Wawrzyniec and Francesco Mazzarini for helpful comments and editorial handling. This study was generously funded through Grants-In-Aid by the New Mexico Geological Society, scholarships by the Department of Earth and Planetary Sciences/University of New Mexico, Graduate Research Development Funds by the Graduate and Professional Students Association/University of New Mexico, and by National Science Foundation (NSF) EAR-0838575 (to Crossey).

1Supplemental File. PDF file containing Figure DR1: 234U evolution diagram; Table DR1: Sample IDs, locations, UTM coordinates, and stratigraphic position; Table DR2: Dated travertine samples used for 234U evolution diagram; Table DR3: Model age calculations (Mesa del Oro); Table DR4: Model age calculations (Mesa Aparejo); Table DR5: Travertine volume calculations; and Table DR6: Dated travertine samples used for age probability histograms. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00946.S1 or the full-text article on www.gsapubs.org to view the Supplemental File.