Eocene–Oligocene chronostratigraphy of ignimbrite flareup volcanic ash beds on the Gulf of Mexico coastal plains

Radiometric dating of volcanic ash is resolving a long-standing problem of obtaining accurate dates for outcropping Cenozoic strata in the northwestern sector of the Gulf of Mexico coastal plains. Strata younger than basal Paleocene Danian Stage contain few fossils capable of providing good age control (Fisher and McGowen, 1969; Fisher et al., 1970; Ewing, 1994). The strata are composed of clastic sediment containing abundant terrestrial plant debris that produced diagenetic conditions resulting in dissolution of most fossils used for biostratigraphic age determination. Reliable fossil-based age control is limited to several major marine incursions (Loutit et al., 1988). Other stratigraphic units are dated by long-distance lithologic correlation of strata from deep-water basinal deposits or from the northeastern sector of the Gulf of Mexico (Loutit et al., 1988; Galloway et al., 2000; Brown and Loucks, 2009). Despite the absence of age-diagnostic marine fossils in the northwestern Gulf of Mexico outcrop belt, these middle Eocene to upper Oligocene strata contain volcanic ash beds that are suitable for radiometric dating (Fig. 1; Table 1).

The radiometric method of age determination remained largely unexploited for Gulf of Mexico coastal plains strata until Heintz et al. (2015) presented ⁴⁰Ar/³⁹Ar dates for a group of east Texas volcanic ash beds. That study also documented the presence of many layers of volcanic ash in upper Eocene and Oligocene strata and made observations on their regional occurrence. It is ironic that many radiometric age determinations are published for detrital zircons in Cenozoic strata of this region (Craddock and Kylander-Clark, 2013; Mackey et al., 2012; Wahl et al., 2016; Xu et al., 2017), but none were presented for ash bed phenocryst zircons. While dating of sanidine phenocrysts yields high-precision results, zircon phenocrysts also provide valuable high-resolution dating for Paleogene volcanic ash. In an area where most dating of formations is inferred and of low accuracy, radiometric dating of ash bed phenocrysts provides a great improvement in age control for outcrop strata that remain the major source of data for regional paleoenvironment, paleogeography, paleodrainage, and paleoclimate reconstructions.

The presence of volcanic ash in Eocene and Oligocene strata in the Texas coastal plains has been noted in many reports (e.g., Bailey, 1926; Renick, 1936; Russell, 1957; Callender and Folk, 1958; McBride et al., 1968; Grigsby and Dennis, 1991), and a few radiometric dates were produced as a byproduct of applied studies (Hurley et al., 1960; Corrigan, 1993; Guillemette and Yancey, 1996) or presented in a Master’s thesis (Deux, 1970). In another part of the Gulf of Mexico coastal plains, Obradovich et al. (1993) reported on some Mississippi volcanic ash beds. The Obradovich et al. (1993) and Guillemette and Yancey (1996) studies presented the first dates using sanidine, obtained by using multiple grains of sanidine phenocrysts to obtain a single numerical date. Heintz et al. (2015) presented the first large study of Gulf of Mexico coastal plains ⁴⁰Ar/³⁹Ar radiometric dating on sanidine single crystals, coupling the geochronology data with major, trace, and rare earth element analysis of bulk ash samples on a large group of volcanic ash beds present in strata of the Brazos River Valley of Texas.

These volcanic ash beds are derived from tephra produced by large volcanic eruptions from the western United States and Mexico (McBride et al., 1968; Heintz et al., 2015). This volcanic activity is related to the ignimbrite flareup across the southwestern portion of North America that occurred as the continent approached the Pacific spreading axis. This generated large amounts of high-silica magma and a corresponding increase in the frequency of large-magnitude silicic volcanic eruptions (Lipman et al., 1972; Ferrari et al., 1999, 2017; McDowell and McIntosh, 2012). In the northern Sierra Madre Occidental (northwestern Mexico), the ignimbrite flareup occurred from 36 Ma to 27 Ma (Ferrari et al., 2017). These eruptions generated ash clouds that spread eastward over the continent to the Gulf of Mexico, depositing ash as far east as the state of Mississippi (Obradovich et al., 1993; Dockery et al., 2017). The Texas sector of the Gulf of Mexico coastal plain extends from 600 to 1200 km from the centers of greatest volcanic activity (McBride et al., 1968; McDowell and McIntosh, 2012). While many ash beds occur as thin layers, meter-thick ash beds (some now altered to bentonite) occur in the section.

Continued dating of Cenozoic Gulf Coast volcanic ash beds expands dating work on the Oligocene Catahoula Group (Fig. 2), a stratigraphic unit containing multiple tephra layers and large amounts of volcanic sediment in south Texas (Bailey, 1926; Renick, 1936; McBride et al., 1968). Sampling is extended to include upper Oligocene strata and expands the study area geographically to include south Texas and the Rio Grande Valley region (Fig. 1). A compilation of ten new ⁴⁰Ar/³⁹Ar-dated volcanic ash beds in the Texas coastal plains is presented in Figure 2. Sampling along the outcrop belt shows that volcanic ash beds are regionally extensive and occur in many horizons within Eocene and Oligocene strata. The recovery of late Oligocene volcanic ash indicates that the northwestern Gulf Coast contains a long-duration record of silicic volcanic activity. Reports of Eocene and Oligocene volcanic ash in the states of Louisiana (Chawner, 1936; Paine and Meyerhoff, 1968; Wrenn et al., 2004) and Mississippi (Obradovich et al., 1993; Dockery et al., 2017) point to regional distribution of volcanic ash beds that are currently dated only in Texas or Mississippi.

Obtaining dates on Catahoula Group volcanic ash is an important advance in dating Gulf Coast Cenozoic strata. Although the Catahoula Group, named and mapped as the Catahoula Tuff (Bailey, 1926; Renick, 1936; McBride et al., 1968), is known to contain large amounts of volcanic ash and glass shards and is considered to be deposited during the time of maximum silicic volcanic activity, the results of previous radiometric dating studies have been equivocal or unsatisfactory. Two attempts made to date outcrop rocks of the Gueydan Formation in the Catahoula Group include K-Ar determination on biotite extracted from “tuffaceous clay” by Deux (1970) and fission-track determinations on apatite extracted from sandstones by Corrigan (1993). Because these dates include some determinations that are much older than expected, they are considered unreliable for dating the time of ash deposition. A more recent report by Craddock and Kylander-Clark (2013) presented a date of 34.07 ± 0.8 Ma on detrital zircons in Catahoula sandstone of western Louisiana, an area where the formation is thin and irregularly preserved (Paine and Meyerhoff, 1968). This date is coeval with latest Eocene volcanic ashes dated by Heintz et al. (2015), but the large uncertainty is consistent with either latest Eocene or early Oligocene deposition and only provides a maximum depositional age. The Catahoula Group dating presented here is consistent with expectations that Catahoula volcanic ash was deposited during peak silicic volcanic activity for western North America.

The high-precision ⁴⁰Ar/³⁹Ar dates presented here on fine-grained, single crystals are made possible by relatively recent developments in multicollector, high sensitivity, ultra-clean mass spectrometers coupled with low-blank extraction systems. Previous ⁴⁰Ar/³⁹Ar dates of coastal plains Texas sanidine were based on analysis of multiple grains (e.g. Obradovich et al., 1993; Guillemette and Yancey, 1996), and not all of these determinations were presented in publication. Single-crystal dating is more accurate than multiple-grain analyses because spuriously young (argon loss) or old (inherited xenocrysts) crystals can be excluded in favor a discrete population of phenocrysts (e.g., Deino and Potts, 1992). Also, for samples that have experienced post-depositional reworking, this method can identify anomalously old detrital grains. In general, the youngest subpopulation of crystal ages is used to determine the age of deposition of the tephra layer. This method of choosing the most accurate age population is also utilized for U-Pb zircon dating. Methodological improvements (e.g., Mattinson, 2005; Schmitz and Schoene, 2007; Condon et al., 2015; McLean et al., 2015) in U-Pb dating by isotope dilution–thermal ionization mass spectrometry (ID-TIMS) have, in a manner analogous to improved Ar-Ar dating, greatly increased the utility of single-grain zircon analyses for high-resolution stratigraphic studies in Cenozoic sections.

The many undated beds of volcanic ash present in the stratigraphic section provide an opportunity to determine a high-resolution chronostratigraphy for Eocene–Oligocene strata throughout the Gulf Coast region, independent of lithostratigraphic correlation–based dating. Radiometric dating provides a direct tie to the paleomagnetic time scale as well. Reliable high-resolution dating is beneficial for all studies involving geologic history, and improves paleogeographic and paleodrainage reconstructions and calculations of rate change of geologic processes. The present study is a contribution of a program started in 2012 with a focus on obtaining radiometric dates and glass geochemistry of single-event volcanic ash beds present in the Texas coastal plains. This program is producing high-resolution dating for Eocene and Oligocene shelf deposits of the northern and western Gulf Coast. This dating can be applied to document a detailed record of cyclic sediment deposition (Yancey and Heintz, 2015) and climate history during a time of cooling and drying in North America (Yancey et al., 2003).

Samples of Gulf Coast volcanic ash are from layers easily identified as being volcanic ash, either by their distinctive white color or by the presence of abundant volcanic glass shards. Layers containing coarse glass shards (i.e., as large as 500 µm) are good candidates for dating provided they also contain sanidine. Excellent candidates for radiometric dating are volcanic ash beds deposited in marine strata or lignites that have planar lower and upper boundaries and show little evidence of post-deposition disturbance. Volcanic ash deposited in nonmarine environments is usually reworked and requires further evaluation to determine acceptability for dating. Although volcanic ash in the Catahoula Group is reworked in all areas examined during this study, samples collected in the Three Rivers area of Live Oak County, Texas, have provided useful dates, including a sample from the large channel fill illustrated by McBride et al. (1968).

Sample collection consists of taking a 5–10 kg sample from an outcrop with preference for material from the base of the volcanic ash bed. In addition to systematic sampling of volcanic ash exposures, the work incorporated samples of opportunity obtained during previous field work. Processing large samples is necessary to recover sufficient material for dating, as the ashes are commonly dominated by volcanic glass. Lab processing, as described by Guillemette and Yancey (1996), removes clay and silt from the sample to produce a sand-size concentrate of minerals. Microscopic examination of the concentrate determines if there is a significant component of phenocryst grains, followed by a decision on whether to attempt radiometric dating on the sample. Samples meeting the criteria for dating were analyzed for ⁴⁰Ar/³⁹Ar geochronology at the New Mexico Geochronological Research Laboratory at New Mexico Tech. Analytical methods are similar to those reported by Heintz et al. (2015). The ⁴⁰Ar/³⁹Ar data are presented in Table 2. Information on irradiation, instrumentation, and analysis is summarized here and provided in the Supplemental Document¹. All age assignments were made without stratigraphic knowledge, thereby eliminating potential bias during data reduction.

Sanidine mineral separation using standard magnetic and heavy liquid methods was conducted at the New Mexico Geochronology Research Laboratory on mineral concentrates provided by Texas A&M University. The crystals were irradiated at the TRIGA reactor in Denver, Colorado, in three separate irradiations that varied from 16 to 32 h (Supplemental Document [^{footnote 1}]) along with Fish Canyon sanidine interlaboratory standard FC-2 with an assigned age of 28.201 Ma (Kuiper et al., 2008). Ages are calculated with a total ⁴⁰K decay constant of 5.463 × 10^–10/a (Min et al., 2000).

After irradiation, at least six crystals of FC-2 from multiple monitor holes were analyzed. Both monitors and sample crystals were fused with a CO₂ laser and the extracted gas was cleaned for 30 s with a SAES NP10 getter operated at 1.6 A and a SAES D50 getter operated at room temperature. The gas was analyzed for argon isotopes using a Thermo Scientific ARGUS VI multicollector mass spectrometer equipped with five Faraday cups and one compact discrete dyonode (CDD) ion counting multiplier. Details of the collector configuration are provided in the Supplemental Table (^{footnote 1}). All data acquisition was accomplished with New Mexico Tech Pychron software (http://www.bgc.org/about_bgc/mission.html), and data reduction used Mass Spec software (ver. 7.875, http://www.bgc.org/about_bgc/mission.html) written by Alan Deino at the Berkeley Geochronology Center in Berkeley, California. Extraction line blank plus mass spectrometer background values are averages of numerous measurements interspersed with the unknown measurements. Blank values are generally small compared to sample gas quantities and are provided for each irradiation in the Supplemental Table (^{footnote 1}).

The preferred eruption ages are generally defined by the youngest group of dates that either form or nearly form a Gaussian distribution as defined by the mean square of weighted deviates (MSWD) value of the selected dates. The reported age is the inverse variance weighted mean of the selected crystals, and the error is the weighted error using the inverse variance as the weighting factor. This error is multiplied by the square root of the MSWD for MSWD > 1. J-factors that monitor neutron dose during irradiation were determined by laser fusion of Fish Canyon sanidine grains and have uncertainties between 0.01 and 0.02%. This uncertainty is propagated into the weighted mean age error that is reported at 2σ precision.

Paleogene high-silica volcanic ash deposits are also amenable to dating with zircon phenocrysts. The sanidine ⁴⁰Ar/³⁹Ar ages are expected to be more precise for Cenozoic rocks, but both of these systems are subject to complications resulting from the incorporation of xenocrystic grains older than the volcanic event producing the ash bed or with the possible loss of radiogenic daughter isotopes from the crystal. Zircon is much more resistant to diagenetic or other secondary alteration than sanidine and can be used for dating where alteration has damaged or destroyed feldspar grains in a sample. The use of both mineral systems from the same sample allows for greater quality control on checking for presumed xenocrystic grains and secondary alteration, permitting a more robust age interpretation. Same-sample ⁴⁰Ar/³⁹Ar sanidine combined with U-Pb zircon dating is presented for four samples to demonstrate agreement of the U-Pb and ⁴⁰Ar/³⁹Ar dating systems within their respective analytical errors. The U-Pb data are presented in Table 3.

Zircon phenocrysts were acquired from the heavy mineral fraction during heavy liquid separation done to concentrate sanidine for ⁴⁰Ar/³⁹Ar dating. Zircons of the South Somerville and Upper Alabama Ferry (Little Brazos) samples obtained from the same aliquot that yielded sanidine phenocrysts sent to New Mexico Tech for ⁴⁰Ar/³⁹Ar dating. Zircons of the Helms and Johnson South samples were obtained from splits of the samples used for ⁴⁰Ar/³⁹Ar dating. Zircon U-Pb analyses were conducted by ID-TIMS in the R. Ken Williams ’45 Radiogenic Isotope Geosciences Laboratory at Texas A&M University. Detailed analytical methods are provided in the Supplemental Document (^{footnote 1}). Most analyses were conducted on relatively large single grains, but, in an attempt to minimize the potential for age biasing by populations of differing grain sizes, aliquots of two to four very small grains were also analyzed.

New ⁴⁰Ar/³⁹Ar and U-Pb dates are presented (Figs. 2–4; Tables 2–3) for groups of volcanic ash beds exposed in Gonzales, Karnes, Live Oak, and Starr Counties. All new Eocene samples come from quarry exposures, and all Oligocene Catahoula Group samples come from outcrop exposures. Samples from Gonzales County were acquired from active bentonite quarries, and samples from Karnes County are from inactive uranium mining pits, whereas Live Oak and Starr County samples are from outcrop exposures. The volcanic ash samples collected in Karnes County can be related to the late Eocene stratigraphy of Eargle et al. (1975), and the Gueydan Formation (Catahoula Group) volcanic ashes are placed in the stratigraphy reported by McBride et al. (1968). Dating of bentonite samples in Gonzales County revealed some conflict with geologic mapping (Proctor et al., 1974), and sample locations are placed on a modified map base (Fig. 5) showing a revised boundary between the Yegua Formation and Jackson Group map units.

A composite stratigraphic section is used for south Texas (Fig. 2), showing formations that can be reasonably identified in that area. The commonly used Caddell and Wellborn Formation names are excluded from this stratigraphic diagram because those two formations are not lithologically distinct from other upper Eocene formations in the area and cannot be reliably identified. Radiometric dating reveals conflict with previous geologic mapping in Gonzales County. Conflicting use of formation names commonly used in reports on upper Eocene strata in Texas show that most of the formation and member names proposed for parts of the upper Eocene section have utility only in a restricted geographic area close to their type sections. When used for geologic mapping, these units are mapped by attempting to laterally trace erosion-resistant sandstone beds that are lithologically similar to beds lower or higher in the section. Consequently, geologic mapping cannot extend the Caddell and Wellborn Formations far along the outcrop belt without using closely spaced drilling work (e.g., Eargle et al., 1975) to determine position of beds buried beneath alluvium fill of large river valleys. This type of control is not available in most areas. Dating of volcanic ash beds can provide the age control needed to correlate strata that otherwise cannot be traced with confidence along the outcrop belt.

Data on maximum grain size and cross-section area of sanidine phenocrysts in the volcanic ash beds are presented in Table 4. Recent work on dating detrital sanidine in sedimentary rocks (e.g., Hereford et al., 2016; Karlstrom et al., 2016; Repasch et al., 2017) is proving to be a valuable tool for determining high-precision maximum depositional ages. Furthermore, these precise detrital crystal ages can often allow individual grains to be linked to caldera sources. Using detrital grain ages for provenance in paleogeographic reconstructions, including river drainage patterns, requires consideration of long-distance airborne travel of tephra, as reported by Sarna-Wojcicki et al. (1981), Walker (1981), and Bonadonna and Houghton (2005). Data presented by Sarna-Wojcicki et al. (1981) show that tephra particle size decreases exponentially with distance, with rapid decrease in particle size in proximal areas and much lower rate of decrease in distal areas. Grain geometry and size also have a significant control on dispersal. It is generally thought that most coarse (i.e., 200–500 μm) sanidine and zircon crystals fall out of the eruptive column within several tens of kilometers (Rose et al., 2003), whereas finer crystals (<100 μm) may travel a couple of hundred kilometers (Matthews et al., 2012). However, some studies have invoked long airbourne transport distances of volcanically sourced detrital zircons and sanidine (Fan et al., 2015; Aslan et al., 2018). As shown by our work here, large (∼500 μm) sanidines can be trasported in the ash column for hundreds of kilometers from their source calderas. All volcanic ash samples dated here are from distal portions of ash beds, and most contain a similar range of tephra particle size.

The coastal plains tephra deposits can be utilized to determine the extent to which coarse material can be transported in the ash plume, as these deposits are ash fall units. Phenocrysts can be reliably separated from other ash components in bentonites, where the glass is altered to clay. In these occurrences, the basal layer contains the coarsest phenocrysts and provides data on the maximum size of phenocrysts carried by the ash plume. Samples of 3–5 kg weight yield sanidine grains up to 150–200 µm in size in deposits of the Eocene–Oligocene volcanic ash beds from east Texas, at sites a minimum of 650 (Brazos County) to 800 km (Houston County) distant from the nearest eruptive center in west Texas. None of the dated ash beds match dates of calderas in the Trans-Pecos area of Texas or the San Juan (Colorado) or Mogollon-Datil (New Mexico) volcanic fields. The major eruptive center of explosive volcanic activity is in northern Mexico, located 300 km west of Texas (McDowell and McIntosh, 2012; Ferrari et al., 2017). More distal deposits of volcanic ash in northern Louisiana (McGuire et al., 2010) contain zircon phenocrysts up to 100 µm in size.

The largest glass shards and phenocrysts occur in the South Somerville ash exposed in Washington County of the Brazos River Valley (Heintz et al., 2015). This ash contains bubble-wall glass shards to 800 µm maximum diameter and sanidine grains to >528 µm maximum axis dimension in a sample of 96 sanidine measured grains (Fig. 6; Table 4). Back-scattered electron element mapping in phenocrysts by microprobe analysis reveals that sanidine is the dominant feldspar mineral and occurs with accessory quartz and plagioclase (Fig. 6A). Minute glass inclusions are present in many grains (Fig. 7), an indication of early, rapid crystallization from the source magma.

These crystal sizes are unexpectedly coarse for distal deposits when compared with compilations of volcanic ash dispersal (Sarna-Wojcicki et al., 1981; Walker, 1981; Bonadonna and Houghton, 2005). The South Somerville ash in east Texas is 650 km distant from the closest silicic eruptive center in the Trans-Pecos area, but the Trans-Pecos does not have a known caldera of the same age (34.10 ± 0.02 Ma) as the South Somerville ash deposit (Heintz et al., 2015). However, the date is coeval with a major episode of silicic magmatic activity in the northern Sierra Madre Occidental (McDowell and McIntosh, 2012; Ferrari et al., 2017). The South Somerville ash also has correspondingly larger pumice clasts, the largest with length of 1–2 mm. Pumice clasts in distal volcanic ash beds usually have larger dimensions than the bubble-wall glass shards. Phenocrysts from the base of the bentonite beds in the Gonzales County bentonite district include some sanidine grains of 150–200 µm diameter, deposited 550 km distant from the closest eruptive center in the Trans-Pecos area.

Samples of four volcanic ash samples were dated by single-crystal methods on co-occurring sanidine and zircon phenocrysts (Fig. 4). The goal of this dating was to evaluate the accuracy of dating with these minerals from the same ash bed and to explore the potential for systematic interlaboratory bias of the type described by Schoene et al. (2006). The expectation was that sanidine dating would provide slightly younger dates than zircon dating, because zircon in these eruptive systems demonstrate nonzero magma residence time and zircon U-Pb dates are commonly slightly older than known eruptive ages or sanidine ⁴⁰Ar/³⁹Ar ages (e. g., Schoene et al., 2006; Deering et al., 2016). This difference highlights the potential for small biases due to sample selection, possibly coupled with differing intermethod dating procedures.

This study shows that ⁴⁰Ar/³⁹Ar sanidine and U-Pb zircon dating of phenocrysts both provide high-resolution dating of Paleogene volcanic ash. The U-Pb dates are slightly less precise than the ⁴⁰Ar/³⁹Ar dates, with uncertainties in the range of ±0.05–0.06 Ma or 0.1%–0.2%. At this level of resolution, the accuracy of some zircon ages may depend on factors not yet resolved. However, for the purposes of correlation and stratigraphic control described here, the equivalence is close enough to justify use of either dating method to produce high-resolution geochronology for volcanic ash, even in samples containing altered feldspars.

The radiometric dates on 18 volcanic ash or bentonite beds from the Texas coastal plains range from 41.84 ± 0.016 Ma to 26.49 ± 0.06 Ma (compiled from dating in south Texas [Fig. 2] and dating in east Texas [Heintz et al., 2015]). Two additional analyses are excluded as they did not yield a reliable young subpopulation of ages, likely related to extensive reworking. Heintz et al. (2015) used bulk ash, glass, and apatite compositions to infer an arc-type volcanic source region within the Sierra Madre Occidental volcanic field of western Mexico for volcanic ash in the Texas coastal plains and adjacent states. The main part of that source area is 1000 km distant from the east Texas part of the Gulf of Mexico coastal plains. Coastal plains Texas volcanic ash and bentonites have composition and geochemical data similar to those of rocks of the ignimbrite flareup (McDowell and Clabaugh, 1979) of the Sierra Madre Occidental, which show pulses of silicic magmatism activity (Ferrari et al., 1999, 2017; Nieto-Samaniego et al., 1999; McDowell and McIntosh, 2012).

The south Texas dated samples correspond to expected stratigraphic order except for dates in the Gonzales bentonite district. Radiometric dating shows that the Johnson South and Helms samples, which are of similar age, are included in strata mapped (Proctor et al., 1974) as being within the top of the Manning Formation in the south part of the map area and within the top of the Yegua Formation to the north of the Guadalupe River (Fig. 4). This is a large difference in stratigraphic level if published mapping is accepted. Quarry exposures also reveal an unreported structural complication: strata in the Helms bentonite quarry have a reversed dip angle compared to strata from surrounding areas. The Gonzales bentonite district lies within the limits of the Balcones-Luling fault trend, with major faulting present at depth in Gonzales County (Tucker, 1967). Movement on Balcones-Luling faults remained active from 70 to 20 Ma (Jackson et al., 2011), but the effects of faulting have not previously been recognized in outcrop mapping. The combination of unreliable formation identification and unrecognized geologic structure is enough to account for the discrepancy found between geologic maps and radiometric dating control.

The new radiometric dates in south-central and south Texas confirm the expectation that volcanic ash beds extend across the state of Texas in the coastal plain outcrop belt and that volcanic eruptive events were episodic. The Plum, Helms, and Johnson South bentonite samples in Gonzales and Fayette Counties have the same date, within error, consistent with being part of the same volcanic ash bed. This date is very close to that of the Graham Road ash of Brazos County (Heintz et al., 2015), clustering together at ca. 35.7–35.8 Ma. Another group of volcanic ash beds, which includes the Conquista, Stoeltje, and Smiley ash beds of Karnes and Gonzales Counties (dates presented here) and the South Somerville, Tarball Quarry, and Gibbons Creek ash beds of Washington and Grimes Counties (Heintz et al., 2015), clusters together at ca. 34.1–34.5 Ma. Each cluster is likely to be from the same source cluster of vents, indicating distinct episodes of volcanic eruption during the Priabonian. The uppermost Eocene cluster of volcanic ash beds extends as far east as Mississippi, where seven bentonites occur in the upper part of the Yazoo Formation (Obradovich et al., 1993). Recognition of pulsing volcanism on a scale of several million years is reported by Ferrari et al. (1999, 2017), and McDowell and McIntosh (2012) reports 1–2 m.y. pulses of ignimbrite emplacement separated by lulls in activity of 1–5 m.y. Grouping of volcanic ash beds reported here occurs at intervals of ∼1 m.y., and the groups were deposited during a time of major ignimbrite emplacement in the northern Sierra Madre Occidental volcanic field (McDowell and McIntosh, 2012; Ferrari et al., 2017). Regardless of whether pulses of ignimbrite emplacement coincide exactly with times of major explosive eruption, variation in rates of eruptive action is observable in the coastal plains deposits and is a useful feature for characterizing eruption frequency in areas with many volcanic ash beds.

Another component of the dating work to pursue is the documentation of the chemistry of glass shards (e.g., Heintz et al., 2015), making tephrochronology a viable tool for correlating Texas Eocene and Oligocene tephra deposits with strata across the Gulf of Mexico region. The new radiometric dates provide numerical age control for the palynology biozones of the late Eocene in Texas (Elsik and Yancey, 2000), and make it possible to integrate the palynology biozones with the paleomagnetic time scale and with the cyclostratigraphy of the Brazos River Valley. There are no published paleomagnetic data for Texas strata of this age, but with a new chronostratigraphic framework, paleomagnetism becomes a valuable tool for dating Eocene–Oligocene episodes of climate change and sea-level change of this region.

The new dates reveal the presence of a thick interval of lower Oligocene strata in the outcrop belt of south Texas. Strata of the lower Gueydan Formation below the Soledad Member provide dates ranging from 32.73 ± 0.08 Ma to 32.46 ± 0.03 Ma, placing them within the lower Rupelian Stage. Brown and Loucks (2009) limited the Catahoula Group in Texas to the late Rupelian and Chattian Stages and concluded that lower Rupelian strata are absent from the outcrop belt across Texas, in line with previous interpretation of a major unconformity being present between the outcropping Jackson and Catahoula Groups (Galloway and Kaiser, 1980; Galloway et al., 1982). The unconformity concept is based on the lack of early Rupelian dates in typical Catahoula strata of the shallow subsurface and outcrop belt, a gap now filled by dates revealing lower Rupelian deposits within the Gueydan Formation.

Field relations also fail to show evidence of a long-duration unconformity in south Texas. The lower Rupelian Gueydan Formation overlies the Frio Clay Formation (of Bailey [1926] and earlier work cited therein), a transitional unit between the upper Eocene Jackson Group and the Oligocene Gueydan Formation. In most places, the Frio Clay Formation has conformable upper and lower contacts or minor disconformable contact with the Gueydan Formation (Bailey, 1926). The Frio Clay Formation pinches out to the north and strata mapped as Catahoula overlie Jackson Group strata with disconformity, but a comparable disconformity is not present in east-central Texas where the uppermost Eocene volcanic ash (34.10 ± 0.02 Ma; Heintz et al., 2015) is conformable with strata mapped as Catahoula Formation (Deussen, 1924). A broad basal Oligocene disconformity is probably limited to the San Marcos uplift of central Texas and thus does not represent a regional event. At the end of the Eocene, there is a Texas-wide change from dominantly marine deposition to dominantly nonmarine deposition. While a local disconformity is an expected feature of this type of environmental change, where radiometric dating is available there is no support for the concept of a statewide regional unconformity.

Instead of a time of erosion, the great thickness of Catahoula Group sediments (Gueyden and Frio Formations) in south Texas (Bailey, 1926) suggests that the high-standing Trans-Pecos volcanic field (Parker et al., 1988) and other highgrounds in the southern Rocky Mountains region built up a nonmarine depositional fan. The Oligocene was a time of resurgent uplift in the southern Rocky Mountains (Galloway et al., 2011; Mackey et al., 2012) and Colorado Plateau (Schulze et al., 2015), forming a western highland source area (Cather et al., 2008; Xu et al., 2017) and drainage system capable of delivering large amounts of sediment to the Gueydan Formation and feeding delta complexes of the subsurface Frio trend in south Texas (Bebout et al., 1978). This suggests that the conglomeratic sediment of the Soledad Member in the Gueydan Formation is a coarse layer that accumulated at the time of maximum buildup. This is a substantially different interpretation of early Oligocene geologic history than that based on presence of a regional basal Oligocene unconformity. Nonetheless, it is based on high-accuracy dating of volcanic ash beds that reveals the presence of a thick interval of strata deposited during the early Rupelian.

New radiometric dating of upper Eocene and Oligocene strata in south Texas provides many numerical dates for strata of the Gulf of Mexico coastal plains and places volcanic ash-bearing intervals into a numerical time scale tied to magnetostratigraphy. Volcanic ash occurs in Eocene–Oligocene strata as far east as Mississippi where they can be tied to marine microfossil biozonation. In Texas, the dating of units with established palynology biozonation and cyclostratigraphy provides a means of integrating lithostratigraphy with the numerical and magnetostratigraphy time scales. Clusters of radiometric dates at ca. 34.1–34.5 Ma and at ca. 35.7–35.8 Ma across the Texas coastal plains indicate times of greater explosive volcanic eruptions and reveal increasing frequency of volcanic ash deposition during the late Eocene. The younger cluster of volcanic ash beds (ca. 34.1–34.5 Ma) extends eastward to Mississippi (Obradovich et al., 1993). Paired ⁴⁰Ar/³⁹Ar dating of sanidine and U-Pb dating of zircon shows that both methods produce comparable high-accuracy dates for Paleogene volcanic ash beds. Sanidine phenocryst grains as large as 528 µm in longest dimension occur in distal volcanic ash deposits as far as 650 km from the eruptive source, and grains as large as 150 µm diameter occur as far as 800 km from eruptive source. Dates of coastal plains volcanic ash beds do not match with dates of known calderas in the Trans-Pecos (west Texas), Mogollon-Datil (New Mexico), or San Juan (Colorado) volcanic fields of the United States, pointing to a source in the northern Sierra Madre Occidental in western Mexico.

Application of new dating improves correlation of strata, reveals unrecognized geologic structures, and makes possible testing of interpretations of geologic history. The presence of lower Rupelian strata in the outcrop belt of south Texas fills in a geologic history for a time interval previously unknown and considered to be missing across the state. Instead of a period of early Oligocene non-deposition, volcanic ash dating provides evidence for the buildup of a nonmarine depositional fan in south Texas. This knowledge greatly improves the accuracy of paleogeography, paleoclimatic reconstruction, and paleodrainage reconstructions for a time interval when Gulf Coast climates changed from warm and wet to seasonal dry (Yancey et al., 2003; Miller et al., 2009).

Support for this study was provided in part by Devon Energy Corporation through a grant supporting graduate research activities at Texas A&M University, and in part by the Department of Geology and Geophysics, Texas A&M University. Charles Smith of BYK Additives (Southern Clay Products) in Gonzales, Texas, Jon Brandt of the Railroad Commission of Texas, Dr. Juan Gonzales and Dr. James Hinthorne of the University of Texas Rio Grande Valley, and Mrs. Johnson of the Johnson Ranch, Three Rivers, Texas, helped in providing access to sample sites. Dr. Youjun Deng of Texas A&M University provided much help in obtaining samples of bentonites. The success of early work done on obtaining radiometric dates is due to the encouragement and support of John Obradovich, U.S. Geological Survey, who continued to conduct research during his retirement years. We also thank our reviewers for help in improving the focus of our study and to acknowledge the considerable help of the Geological Society of America editors in shaping the clarity of our writing and correcting oversights that authors can make on the need for explanation when they are completely immersed in a subject.