Abstract

The ∼400,000 km2 calc-alkalic Sierra Madre Occidental volcanic field is the largest component of the Tertiary ignimbrite flare-up of western North America. Laser-fusion 40Ar/39Ar geochronology of sanidine and anorthoclase from ignimbrites sampled within three areas of concentrated geologic mapping in the Mexican states of Chihuahua and Durango provides the first examination of fine-scale temporal fluctuations within this significant area of the ignimbrite flare-up. Overall, the 112 40Ar/39Ar ages range from 28 to 46 Ma, with a separate group near 24 Ma. The oldest ages (40–46 Ma, with a strong cluster at 44–46 Ma) are found only along the eastern margin of the field. The predominant age concentration (28–36 Ma) is found across two east-west traverses of the volcanic field that are separated by 500 km. Within this interval, there are strong concentrations at 28–30 Ma, ca. 33 Ma, and 35–36 Ma in the northern traverse, and at 30–31 Ma and at 31.5 Ma in the southern traverse. A cluster of ages near 24 Ma is present in the western part of both traverses. Ignimbrite sequences of similar age are oriented in belts generally parallel to the axis of the Sierra Madre Occidental volcanic field. The belts partly overlap but are progressively younger toward the west. This geographic pattern fits with numerous studies that relate the position of arc magmatism to the delamination/rollback of the subducted Farallon slab. However, a more difficult challenge is to reconcile a pattern of discrete, brief (∼2 m.y.) and intense episodes of magmatism to behavior of the Farallon plate.

INTRODUCTION

Tertiary volcanic rocks of the ignimbrite-dominated Sierra Madre Occidental volcanic field characterize the geologic landscape of western Mexico. The Sierra Madre Occidental volcanic field, the largest contiguous exposure of Tertiary volcanic rocks in North America, extends in a continuous, broad arc from the United States border 1200 km southward to the city of Guadalajara (Fig. 1). With an original area estimated at 400,000 km2, this field is approximately five times the size of similar and broadly contemporaneous volcanic fields in the western United States (inset Fig. 1), which together record the ignimbrite flare-up of western North America (e.g., Henry et al., 2010). As the term suggests, impressive stacks of large-volume ignimbrites constitute a record of major pyroclastic eruptions from scattered large caldera sources. For the Sierra Madre Occidental volcanic field, accumulated ages, initially from K-Ar dating (e.g., McDowell and Mauger, 1994), and later from 40Ar/39Ar geochronology, indicate that these eruptions occurred mostly between 46 and 28 Ma, with only localized activity outside of these times. This active period of volcanism coincided with a very quiet tectonic interval separating Late Cretaceous–Early Tertiary shortening related to the Laramide orogeny from Neogene extensional deformation characteristic of the Basin and Range region. It is of interest therefore to understand fully the patterns and rates of volcanic activity, the mineralogical and chemical composition of the volcanic rocks, and ultimately to explore the root causes of the ignimbrite flare-up.

For the 20% of the flare-up area within the United States, a long history of field-based studies has fed into detailed examinations of the chronology and composition of the rocks. In contrast, the flare-up within Mexico has received far less attention over the years, and accumulation of data has preceded mapping in some cases. Therefore, much less is known about caldera sources and the number of ignimbrites that have erupted from them. It will be many years before the level of field study in Mexico is comparable to that in the United States.

This paper contains the first comprehensive results of 40Ar/39Ar dating from areas within the northern and central parts of the Sierra Madre Occidental volcanic field. We examine in detail the history of volcanism in this region by focusing upon precise eruption ages for major units from three contiguous areas of geologic studies (Fig. 2). Results obtained from widely distributed samples that lack geologic context have not been included here. Rather, we present data for samples linked to well-documented stratigraphic successions produced from mapping projects concentrated in three transects (Figs. 1 and 2). The field studies in these transects, which are described in the following section, have been completed by graduate students at The University of Texas at Austin and graduate students and faculty from East Carolina University.

GEOLOGIC CONTEXT

Volcanism during the Eocene–Oligocene ignimbrite flare-up was preceded by Late Cretaceous and Early Tertiary arc magmatism related to subduction of the Farallon plate. In western Mexico, intrusions related to this earlier (Laramide) arc are exposed broadly across both Sonora and Sinaloa States. Initial geochronology, mostly based upon K-Ar dating of biotite and hornblende from plutons, generally indicates eastward younging of magmatism interpreted to reflect the pattern of Farallon plate subduction (Keith, 1978; Damon et al., 1981, 1983; Henry et al., 2003). Recently, more robust 40Ar/39Ar and U-Pb dating of both volcanic and intrusive rocks of the Laramide arc has provided ages as old as 90 Ma in eastern Sonora, complicating interpretation of both the age pattern and its tectonic significance (McDowell et al., 2001; Valencia-Moreno et al., 2006; Roldán-Quintana et al., 2009). The mid-Tertiary ignimbrite plateau of the Sierra Madre Occidental volcanic field conceals the eastward continuation of the older arc.

Exposures of volcanic components of the Laramide arc are limited to easternmost Sonora and Sinaloa. In southern Sinaloa, Henry and Fredrikson (1987) mapped a widespread and thick sequence of heavily altered volcanic units of intermediate and felsic composition that they called the lower volcanic complex. They described a heterogeneous map unit that includes pyroclastic falls and ash-flow tuffs, felsic and intermediate lavas, volcaniclastic sediments, and minor hypabyssal intrusions. A ubiquitous, heavy propylitic alteration has precluded direct dating, but contact field relations and the presence of plutonic boulders in volcanic agglomerates indicate that the lower volcanic complex is essentially coeval with the Laramide plutonic episode (Henry and Fredrikson, 1987; Henry et al., 2003). In eastern Sonora, U-Pb zircon dating of the equivalent volcanic rocks has provided ages of 70 and 90 Ma (McDowell et al., 2001).

Henry and Fredrikson (1987) also recognized a sequence of younger, less-altered intermediate composition lava flows, agglomerates, and red beds that conformably underlie the ignimbrites in southern Sinaloa. Aranda-Gómez et al. (1997) described gravels within this sequence that contain clasts of the lower volcanic complex in exposures along the Durango-Mazatlán highway, ∼10 km south of El Palmito (area 3 of Fig. 2). Equivalent units have not been recognized beneath the ignimbrites in eastern Sonora. However, at Tomochic, Chihuahua (area 1 of Fig. 2), relatively unaltered andesitic lavas are intercalated with the earliest ignimbrites (Swanson and McDowell, 1985; Wark et al., 1990). A geochemical study by Wark (1991) demonstrated a plausible petrogenetic relationship by fractional crystallization of the andesites and the locally derived ignimbrites at Tomochic. It is likely that intermediate composition lavas represent an early phase of Tertiary volcanism throughout much of the Sierra Madre Occidental volcanic field. However, farther to the east, the ignimbrites lie directly and unconformably upon folded mid-Cretaceous limestone units, and the basal intermediate lavas are apparently absent.

In many areas within the northern Sierra Madre Occidental volcanic field, mafic lavas generally of basaltic andesite composition overlie the ignimbrites. Typically, this relationship is conformable and lacks intervening clastic sedimentary units. In a reconnaissance study of these rocks in central and eastern Chihuahua, Cameron et al. (1980, 1989) noted their similar composition to widespread lava units throughout the western United States, including those of the Columbia River basalt series. They assigned these rocks to a regional suite that they called Southern Cordilleran basaltic andesites (SCORBA). Cameron et al. (1989) proposed that the basaltic andesites reflected a shift from a compressional tectonic setting associated with the rhyolitic volcanic arc to tensional conditions associated with the initiation of Basin and Range extension in Chihuahua. However, K-Ar ages for the basaltic andesites within mapped areas of the Sierra Madre Occidental volcanic field are only slightly younger than ages for the underlying ignimbrite units (McDowell, 2007, and mapping compiled therein). Furthermore, the age distributions of both rock types show the same geographic variation. The petrogenetic study of Wark (1991) on the rocks adjacent to the Tomochic caldera complex (area 1 of Fig. 2) established a plausible genetic link among the older andesites, the ignimbrites erupted from the caldera, and the capping basaltic andesites in a crystal fractionation scheme that can be related to the evolution of an individual magma chamber. Near Nazas, Durango (area 3 of Fig. 2), Aguirre-Díaz and McDowell (1993) showed that there are distinct compositional differences between these capping basaltic andesites (ca. 30 Ma) and younger basalts (24 Ma) that are clearly related stratigraphically and geochemically to Basin and Range extension.

Finally, a group of peralkaline rhyolitic ignimbrites distinctly different from the regional calc-alkaline suite occurs within a radius of ∼100 km from Chihuahua City (area 1 of Fig. 2). Cameron et al. (1980) examined some rocks of this suite in their reconnaissance to the southeast of that city. Although the bulk composition of these rocks is only borderline peralkaline, the devitrified groundmass displays rosettes of aegerine and Na-Fe amphiboles. They were named “ferroaugite rhyolites” by Cameron et al. (1980) because of the presence of iron-rich mafic phenocrysts preserved in fresh samples. Similar peralkaline rocks have been mapped extensively in areas to the north of Chihuahua City (area 2 of Fig. 2) by Mauger (1981, 1992), and to the west of Chihuahua City in areas compiled by McDowell (2007). The 40Ar/39Ar sanidine ages for these rocks are presented and discussed in this paper, although the composition and origin of these rocks are not considered further here.

AREAS OF SAMPLING

Area 1 (Fig. 2) is a 250-km-long transect at ∼28°N across the Sierra Madre Occidental volcanic field that has recently been compiled based upon mapping by 11 graduate students at the University of Texas at Austin (McDowell, 2007). The stratigraphic units defined in original mapping were grouped into a regionally coherent stratigraphy appropriate for the publication scale of 1:250,000. The compilation includes the original geochemistry and K-Ar geochronology, as well as more recently obtained data that include the 40Ar/39Ar ages presented here. The western half of area 1 crosses the unextended Sierra Madre Occidental volcanic plateau beginning in Yécora, Sonora, and the eastern half continues across tilted fault blocks formed during postvolcanic Basin and Range faulting to near Chihuahua City (Fig. 2). Ten calderas have been documented or suggested from mapping within or near area 1 (Table 1; Fig. 2). Four of these (calderas 1D through 1G) were described in a separate study conducted to the south of Tomochic (Fig. 2) (Swanson et al., 2006). At least one of those four produced an ignimbrite (San Filipe tuff) that has been mapped and sampled within area 1. Caldera 1J is just east of area 1 and was mapped by Megaw (1990).

Area 2 combines a group of mapped areas extending from Chihuahua City northward for ∼170 km between longitude 106°W and 107°W (Fig. 2). Students and faculty at East Carolina University conducted most of this mapping (McDowell and Mauger, 1994, and references therein). Students of the University of Texas at Austin mapped the Sierra Gallego area, which is located at the northernmost extent of area 2 (Keller et al., 1982). Area 2 follows the easternmost extent of the Sierra Madre Occidental volcanic field, where volcanic rocks are ponded against the high-amplitude folds of the Chihuahua tectonic belt.

McDowell and Mauger (1994) summarized the general stratigraphic relationships and ages along area 2, and in mapped areas in the eastern part of area 1. A prominent feature of the stratigraphy in area 2 is the occurrence of three widespread peralkaline (fayalite and Na-Fe pyroxene-bearing) ignimbrite units that dominate the upper portion of the volcanic section throughout most of this area (Mauger, 1981). These rocks correspond to the ferroaugite rhyolites described in a reconnaissance study located nearby by Cameron et al. (1980). Four calderas within area 2 (Table 1) include the well-documented Majalca Canyon caldera (2C; Mauger, 1992), and probable sources (calderas 2A and 2B) for the peralkaline units, where areas of shallow subsidence bounded by minor faults have been identified (Mauger, 1992; Mauger, 2011, personal commun.). A fourth volcanic source (2D) in the northern part of area 2 is a 25-km-diameter center that has produced several large felsic and intermediate composition lava flows along with lesser-volume ignimbrites (Gallego rhyolites and related rocks of Keller et al., 1982; Mauger, 1981, 1988). This center is not a caldera.

Area 3 is located in the state of Durango, where much of the earliest mapping by the University of Texas at Austin students was conducted. The westernmost part of this transect follows Mexican National Highway 40 west from the city of Durango in a west-southwest direction (Fig. 2). Mapping and K-Ar chronology along this leg of area 3 were described in McDowell and Keizer (1977) and Swanson et al. (1978). An eastern leg extends area 3 to the NNE from Durango City to include later studies in the Nazas-Rodeo area of northeastern Durango State (Aguirre-Díaz and McDowell, 1991, 1993; Luhr et al., 2001). The total combined length of area 3 is 240 km. Mapping is discontinuous along the NNE segment. Geology along both legs of this transect has been described in a field trip guidebook for a meeting of the International Association of Volcanology and Chemistry of the Earth’s Interior (Aranda-Gómez et al., 1997). Calderas identified within area 3 include the Chupaderos caldera complex (caldera 3A), just north of Durango City, a well-documented source for two important ignimbrite units there (Swanson et al., 1978), and another (caldera 3B) that was proposed, but not yet identified, for the tightly conformable sequence of 24 Ma ignimbrite units prominently exposed to the west of El Salto (Table 1; Fig. 2).

APPROACH TO GEOCHRONOLOGY

The geochronology presented here is based upon samples collected and prepared for K-Ar dating during prior field studies. More precise 40Ar/39Ar ages were obtained from sanidine or anorthoclase phenocrysts. Where possible, we prepared new mineral separates from coarsely crushed material to enable single-grain measurements. If coarse material was no longer available, we conducted multigrain measurements using aliquots of ∼50 grains each from the original finer separates of feldspar. Single-grain 40Ar/39Ar measurements allow the statistical elimination of spurious ages from incorporated xenocrysts and antecrysts or from altered grains. However, the multigrain analyses are somewhat inferior for this purpose. The 40Ar/39Ar ages of biotite or plagioclase phenocrysts were not measured, as results from those minerals may not record cooling at the time of eruption (Bachmann et al., 2007), but rather prolonged crystallization within the magma chamber and/or the incorporation of undegassed xenocrysts from slightly older volcanic strata. However, some K-Ar ages previously obtained for biotites and/or plagioclases, and some U-Pb ages from zircon have been added to provide more complete age coverage in parts of areas 1 and 3, where few ignimbrites contain alkali feldspar phenocrysts (McDowell and Mauger, 1994; McDowell, 2007). The K-Ar and U-Pb analyses yielded higher uncertainties that obscure small differences in age, such as those observed between coeval sanidine and biotite.

METHODS

The coarse mineral separates were prepared from crushed and sieved materials by magnetic separation, ultrasonic cleaning in 10% HF, heavy-liquid techniques, and hand picking. For the finer-grained separates, a precise density separation was used to remove remaining quartz and plagioclase. All 40Ar/39Ar ages were analyzed at the New Mexico Geochronology Research Laboratory using methods similar to those detailed in McIntosh and Chamberlin (1994). Feldspar separates and neutron flux monitors (using an age of 28.02 Ma for Fish Canyon Tuff sanidine; Renne et al., 1998) were irradiated in several different irradiation batches in stacked, machined aluminum trays for 7–15 h in the D-3 position of the Nuclear Science Center reactor at Texas A&M University. Single crystals of the flux monitor and single- and multicrystal aliquots of samples were fused by a CO2 laser and analyzed using an MAP 215–50 mass spectrometer connected to an automated extraction line. More detailed analytical information, including gettering, sensitivity, blanks, correction factors, and error calculation methods, is given in the footnotes to Supplemental Tables 11 and 22.

Given sufficient material, at least 10 individual laser-fusion analyses were conducted for each sample. The feldspar analysis provides a K/Ca ratio, which is calculated from the ratio of 39Ar produced from K to 37Ar produced from Ca during sample irradiation. A K/Ca value of 10 was arbitrarily chosen to distinguish sanidine (higher K/Ca) from anorthoclase (lower K/Ca). Results with K/Ca ≤ 0.5 were assumed to be analyses of plagioclase phenocrysts and were excluded from the age calculations. Also discarded were analyses showing abnormally low radiogenic argon contents (assumed to be due to quartz or plagioclase) and those with higher than typical atmospheric argon corrections (usually due to gaseous inclusions, adhering matrix, or altered grains). This filtering required supplemental analyses for some samples with unfavorable phenocryst mineralogy. Remaining statistical outliers were also eliminated. Cumulative statistical analysis of the acceptable individual experiments was employed to obtain an age for each sample, which was based upon a weighted mean age and uncertainty for the unimodal distribution.

40Ar/39Ar GEOCHRONOLOGY

The 112 40Ar/39Ar ages are summarized in Table 2, where they are arranged geographically, from west to east for areas 1 and 3 and from south to north for area 2. The ages are given cardinal numbers in the left-most column of the table for easier reference. Ages and analytical data for 50 of the samples have been published previously (Luhr et al., 2001; McDowell et al., 2005; McDowell, 2007). Small numbers of the previously published ages have been slightly revised, as noted in Table 2, either to include additional analyses since publication, or because the age used for the irradiation flux monitor has been changed (see revised supporting data in Supplemental Table 1 [see footnote 1]). All results from area 2, along with ages for three samples from the west of area 1, and 23 of 31 analyses from area 3 are reported here for the first time. Supporting analytical data tables for these are in given in Supplemental Table 2 (see footnote 2). Corresponding analytical and graphical details for revised and newly published ages are shown in Supplemental Figures 13 and 24, respectively.

Figures 3A, 3B, and 3C present cumulative frequency plots and relative probability plots for the samples from each area. In area 2 (Fig. 3B), peralkaline rocks were sampled in considerable detail; of the 12 ages indicated in the other tables and figures, only six, chosen to display the full age range, have been plotted in Figure 3B. Figures 4A, 4B, and 4C, display the sample locations and a geographical projection of the ages separately for all three areas. Colors and symbols identify specific groups of samples discussed together in the following sections.

DISCUSSION

Data Limitations and Remedies

Although the sampling for this study represents less than 20% of the total area of the Sierra Madre Occidental volcanic field, each area provides a different regional perspective of the patterns of volcanism (Fig. 2). Area 1 is oriented orthogonally to the major axis of the field and nearly covers its full width, crossing both the undisturbed western portion and an eastern part that is disrupted by Basin and Range–style faulting. In contrast, area 2 provides an arc-parallel perspective along the easternmost margin of the Sierra Madre Occidental volcanic field in Chihuahua. Area 3, located ∼300 km farther to the south in the state of Durango, combines features of the first two. Its western section traverses the volcanic field orthogonally, and its eastern portion obliquely crosses the eastern portion of the field, including areas that have experienced later faulting and tilting. Although additional coverage would be desirable, we believe that the timing and spatial distribution of volcanism within these three areas can provide a first approximation that is applicable to the entire central and northern parts of the Sierra Madre Occidental volcanic field.

Even within the three areas, however, there are limitations due to an uneven distribution of sampling (see Figs. 4A, 4B, and 4C). Typically, samples were collected after the stratigraphy within each map area was generally understood but before correlations were made to adjacent areas. Limited exposures of the base of the volcanic section precluded complete stratigraphic sampling within the nonextended western portions of areas 1 and 3. Another significant problem is the regional variation in phenocryst mineralogy favorable for 40Ar/39Ar geochronology. Toward the eastern side of the Sierra Madre Occidental volcanic field, virtually all felsic ignimbrites carry phenocrysts of sanidine or anorthoclase. Toward the west, however, alkali feldspars become less common, and the phenocryst cargo of most ignimbrites is dominated by quartz, plagioclase, and biotite. As a result, some major units in the western Sierra Madre Occidental volcanic field are not represented in the data set of sanidine and anorthoclase 40Ar/39Ar ages. To supplement information for some sparsely sampled areas, we have included some less precise K-Ar ages from biotite and plagioclase, and U-Pb zircon ages to the plots of Figure 4. Specifically, in Figure 4A, we have included published K-Ar ages near 24 Ma on biotite and/or plagioclase from several units from the western part of area 1, some K-Ar ages near 30 Ma from the central part (McDowell, 2007), and some U-Pb zircon ages from older tuffs in the eastern part of area 1 (McDowell and Mauger, 1994). In Figure 4C, we show K-Ar ages of 28–33 Ma for sanidine and anorthoclase (McDowell, 1980, personal observ.) from units sampled within the unmapped gap of area 3. In contrast to the situation discussed earlier for 40Ar/39Ar ages, the larger analytical uncertainties of these K-Ar and U-Pb ages mask the problem of their interpretation.

Tests of Correlation

Because only small numbers of caldera sources for specific ignimbrites have been identified in the Sierra Madre Occidental volcanic field (Table 1), it is impossible to place most dated samples at the sites of their eruption. Although many of the mapped ignimbrites probably traveled modest distances (e.g., <100 km), it is possible that the most voluminous may have traveled 150 km or more, and may have been sampled repeatedly in separate map areas as presumed different units. Sampling density is uneven and especially concentrated in areas near known calderas. There has been concentrated sampling near Durango City (area 3) near the Chupaderos caldera complex (Swanson et al., 1978; McDowell et al., 2005), and in the central part of area 1 near the Tomochic caldera complex (Swanson and McDowell, 1985; Wark et al., 1990; McDowell, 2007). Peralkaline tuffs in area 2 and the eastern area of area 1 were thoroughly sampled because of their unusual composition and widespread distribution throughout an area that had undergone major postvolcanic extension (McDowell and Mauger, 1994).

A statistical test can be applied to groups of samples that are within reasonable geographic proximity and have similar ages to examine whether they might be from the same eruptive unit. The distinction between “same eruptive unit” and “same map unit” (which may include more than one eruptive unit) should be kept in mind for this exercise. Because equivalent age does not prove correlation, other qualitative factors require consideration. Ages and uncertainties, and measured K/Ca values and uncertainties, along with experiment type (single or multicrystal), and qualitative phenocryst petrography (only if available for the specific samples dated) are tabulated for nine groups of samples tested (Table 3). Although phenocryst proportions within a single unit may vary somewhat with distance from its eruptive source, profound differences can negate correlations suggested by these other parameters. The final columns of Table 3 provide the weighted mean 40Ar/39Ar age, standard error, and mean square of weighted deviates (MSWD) for each specific group. The following discussion follows the sequence of groups, and the results are illustrated graphically in the six panels of Figure 5.

The first two groups in Table 3 provide a clear example of the approach. Listed first are six samples, two of which (Table 2, entries 8 and 17) have been correlated in the field as Vista tuff, and erupted from the nearby Las Varas caldera (Wark et al., 1990). These are listed together with four uncorrelated samples (1, 3, 4, and 6) having closely similar ages and petrography that were collected near the caldera and up to 150 km toward the west. The weighted mean age for the six samples, 33.37 ± 0.04 Ma (Fig.5A), along with the tight agreement of the K/Ca values, suggests that all of these samples are of the phenocryst-rich Vista tuff, in which potassium feldspar is present but subordinate. Sanidine-anorthoclase phenocrysts are rare in all other tuffs sampled in the western part of area 1.

The second group in Table 3 includes eight samples that have broadly similar ages to that of the Vista tuff. One of the samples (16) in this group is petrographically different from the Vista tuff, although its age overlaps that of the Vista group within uncertainty (Fig. 5A). This sample is from a basal unit of the younger (ca. 31.7 Ma) Rio Verde tuff, which is described as lacking potassium feldspar and erupted from the Tomochic caldera, which significantly overlaps the Las Varas caldera (Swanson and McDowell, 1985). The data line in Table 2 for sample 16 indicates that only two of 30 crystals analyzed were sanidine. McDowell (2007) contains the full data block for this sample. It appears that these analyzed sanidines are xenocrysts picked up from the Vista tuff. This example shows the value of including petrography as a qualitative check on these correlation tests. Two other samples analyzed from the Rio Verde tuff (see Table 2, entries 12 and 14) are both from higher flow units within the Rio Verde tuff that contain small quantities of potassium feldspar. They indicate its younger 32.0 Ma age (Table 2).

The remaining samples in the second group (Table 2, entries 27, 30, 31, 33, 43, 44, and 45) are all from the eastern part of area 1. The mean age of this group appears to be close to that of the Vista samples, but the individual ages scatter more broadly (Fig. 5A). Several of these results are from multigrain analyses. The K/Ca ratios are all markedly lower than those for the Vista samples. None of these eastern samples appears to be of the Vista tuff. In fact, it is likely that multiple sources produced these tuffs, although subsets within the group may be more closely related.

Three groups representing separate map units of peralkaline rocks occur in the eastern part of area 1 (mapped as Frijol tuff) and throughout much of area 2 (mapped as Acantilado and Campana tuffs [Mauger, 1981]). Mauger (2011, personal commun.) suggested that the Acantilado and Campana tuffs (as well as the massive, aphyric Cryptic tuff, which occurs stratigraphically between them) all were derived from the same general area (Fig. 2; Table 1; calderas 2A and 2B). The mapped outcrop area of the Frijol tuff is ∼75 km to the southwest of this suggested source (McDowell, 2007). The sanidines of the peralkaline rocks have low Ca concentrations and thus variable K/Ca values that limit the usefulness of this parameter. In addition, multigrain analyses were used for most of the samples (Table 3). Nevertheless, ranges for K/Ca are distinct for the three groups, and they appear to have close but significantly different ages (Fig. 5B). The age for the geographically separate Frijol tuff also falls between those of the Acantilado and Campana tuffs, perhaps reflecting a source complex that was active for ∼400,000 yr.

The Gallego rhyolites (Table 3; Fig. 5C) comprise large-volume felsic to intermediate flows, along with a lesser volume of ignimbrites found within an area ∼50 km in diameter in the northern part of area 2 (Mauger, 1988; Keller et al., 1982). The rocks were erupted from local sources rather than a single caldera. Nevertheless, the ages are tightly distributed at 37.87 ± 0.10 Ma, and the K/Ca ratios are uniform with one exception. The Gallego rhyolites represent an unusual eruptive setting for the Sierra Madre Occidental volcanic field, but they are similar in lithology and volume to flows studied in the Trans-Pecos volcanic field of Texas, including the Bracks and Star Mountain rhyolites, and the Crossen trachyte. The Texas rocks are ∼1 m.y. younger than the Gallego rhyolites (Henry et al., 1994).

The southern part of area 2 contains a number of ignimbrites comprising several map units with ages near 45 Ma (Table 3; Fig. 5D). Many of these are located near the Majalca Canyon caldera (Mauger, 1992; Table 1; Fig. 2). A group of nine dated samples are arranged by decreasing age in Table 3. The six oldest ages form a tight statistical cluster, suggesting that the samples could be derived from the same source, although the K/Ca ratios vary somewhat. Including the youngest three ages, the results show a spread in ages and in K/Ca ratios that probably indicate eruptions from more than one source. No specific petrographic data are available for the dated samples.

The remaining two groups of data in Table 3 are from area 3. The conformable upper sequence of ignimbrites studied from the Durango area was tested (McDowell and Keizer, 1977), and results are shown in stratigraphic order in Table 3. Six of the seven results are within a range of 0.5 m.y., but the weighted mean ages and uncertainties are too large to indicate a single time of eruption (Fig. 5E). The necessity to use multigrain experiments limits the utility of the K/Ca ratios for most of these samples. The four El Salto tuffs have ages and K/Ca values suggesting a single source (Table 3; Fig. 5F), although no caldera structure has been mapped within this area of difficult access.

Time Patterns

Complete sets of 40Ar/39Ar ages are plotted separately for each area in Figures 3A, 3B, and 3C as cumulative frequency plots (top) and relative probability plots (bottom). The predominant temporal pattern for all three areas is of sharp peaks separated by broader gaps. Some of the peak heights have been exaggerated by multiple samples from a specific source, and some gaps may narrow or disappear with more comprehensive regional sampling. However, within the overall range from 46 to 24 Ma, it appears that the intensity of eruptive activity has fluctuated significantly with typical variations of a few million years.

Across area 1, 40Ar/39Ar ages range from 28 to 46 Ma, with a significant gap between 30 and 31.5 Ma and only sparse ages older than 36.5 Ma (Fig. 3A). Within the most active period, from 31.5 to 36 Ma, two prominent clusters can be discerned. The oldest is 35–35.5 Ma; a stronger one occurs at 33–33.5 Ma, roughly coincident with the Eocene-Oligocene time boundary. This peak is enhanced by ages for several samples of the Vista tuff, but there are also numerous similar ages for tuffs unrelated to the Vista tuff from the eastern part of area 1 (Table 3). The peak at 29–30 Ma consists dominantly of ages for peralkaline rocks from the eastern portion of area 1. However, major calc-alkaline ignimbrites with ages of 30 Ma have been noted in the Divisadero area, 85 km south of Tomochic (Swanson et al., 2006). The youngest 40Ar/39Ar ages found within area 1 are 28–29 Ma, from ignimbrites and rhyolitic plugs associated with eruption from the Basaseachic caldera and during final stages of resurgence at the Tomochic caldera complex.

Ages for area 2 range from 29 to 46 Ma, but those younger than 35 Ma are almost entirely from peralkaline rocks (Fig. 3B). There are significant gaps in volcanic activity between 37 and 30 Ma (occupied by one age), and between 44 Ma and 38.5 Ma (occupied by two ages). The intervals of strongest activity are between 46 and 44 Ma, mainly the Majalca Canyon tuff units, and centered on 38 Ma, the latter mostly from the geographically restricted Gallego rhyolites. The peak of ages at 29–30 Ma, from peralkaline rocks, is virtually coincident with those for the peralkaline rocks in area 1. To reduce the effect of this peak, only one-half of the measured ages have been plotted in Figure 3B.

In area 3, the 40Ar/39Ar ages range from 24 to 44 Ma and cluster into three brief episodes at ca. 24, 30–32.5, and 42–43 Ma, with no dated rocks in between (Fig. 3C). For the middle episode, two closely spaced peaks can be discerned, one at 31.5 Ma, which includes rocks clearly erupted from the Chupaderos caldera complex north of Durango City (Fig. 2; Swanson et al., 1978), and a slightly younger peak at 29.5–30.7 Ma, which corresponds to a younger sequence of conformable strata just to the west of Durango City. These younger strata were not derived from the Chupaderos caldera. Dated rocks from the Nazas-Rodeo area, ∼110 km to the north-northeast of Durango City (Fig. 2), also contribute to the ca. 30 Ma peak, but include ages of 42–43 Ma. The youngest cluster of ages at 24 Ma in area 3 is associated with a conformable sequence near El Salto and to the west.

Geographic Patterns

The upper portions of Figures 4A, 4B, and 4C show the geographic distribution of the sample locations and their 40Ar/39Ar ages for the three areas. The figures illustrate areas where sample locations have been clustered. The most extreme case is near Durango City (Fig. 4C), where the transect map is expanded to show the ages adjacent to their sampling points. Elsewhere, a concentration north of Chihuahua City (Fig. 4B) is dominated by samples with ca. 45 Ma ages, mostly from the Majalca Canyon area (Table 3). Also evident are areas where sampling is sparse or absent, such as west of Durango City, where few of the tuffs carry potassium feldspar phenocrysts, and north of Durango City, where there is a gap in geologic map coverage (Fig. 4C). Another sparsely sampled area is to the west of Cuauhtémoc in area 1 (Fig. 4A), where the calc-alkaline units lacked phenocrysts of sanidine and/or anorthoclase.

The bottom portions of the three parts of Figure 4 show the ages projected onto a line parallel to the orientation of the transects, including the jog in direction of area 3. Note that some K-Ar plagioclase and biotite ages and some U-Pb zircon ages with inferior precision have been added to Figures 4A and 4C, to show the full distribution of known ages for ignimbrites.

The oldest ages, those between 44 and 46 Ma, are located primarily in area 2, with a strong concentration north of Chihuahua City (Fig. 4B). Ignimbrites with ages above 40 Ma occur sparsely farther north in area 2, in the easternmost part of area 1 within 100 km of Chihuahua City (shown primarily by the U-Pb ages in Fig. 4A), and in the northeastern part of area 3 (Fig. 4C). All of these locations are along the eastern margin of the Sierra Madre Occidental volcanic field, where the basal ignimbrites pinch out against preexisting folds produced during the Laramide inversion of the Mesozoic strata of the Chihuahua basin. It is possible that ignimbrites of this oldest age interval may be concealed beneath younger volcanic rocks for some distance to the west in area 1, where the base of the volcanic section has not been observed. In the northern portion of area 2 (Fig. 4B), there is a complex of intercalated rhyolitic plugs, flows, and ignimbrites (Gallego rhyolites) that is ca. 38 Ma in age (Fig. 4B). This age and eruptive style appear to be unique for the areas that we have studied within the northern and central Sierra Madre Occidental volcanic field.

Beginning at ca. 36 Ma, ignimbrite eruptions became more intense and widespread in area 1 (Fig. 4A). There is a concentration of ages between 36 and 35 Ma throughout the eastern 100 km of the area. Scattered ages obtained from near Tomochic and west of Yécora indicate that activity as old as 36 Ma could be present but unexposed across the entire width of the Sierra Madre Occidental. Certainly, ignimbrites of age between 33.9 and 32.4 Ma can be found across the entire width of the Sierra Madre Occidental from south of Chihuahua City to the western margin of area 1 (Fig. 4A). However, between Tomochic and Yécora, these are represented only by the Vista tuff. Rocks of this age are absent from area 2 and also possibly from area 3.

The time interval between 31.5 and 28 Ma was exceedingly magmatically productive in all three areas. Eruptions occurred at ca. 31.5 Ma from the two best-studied caldera complexes: both the Santuario and Aguila tuffs from the Chupaderos caldera complex near Durango City (Fig.4C), and the Rio Verde tuff, the second ignimbrite erupted from the Tomochic caldera complex (Fig. 4A). West of Tomochic, the Basaseachic caldera erupted a major tuff at 29 Ma. Scattered activity as young as 28 Ma also occurred throughout the western portion of area 1, including minor final activity related to the Tomochic caldera. Similarly, in area 3, ignimbrites of age 33–28 Ma are widespread throughout the eastern part, where they are represented by numerous K-Ar ages obtained in unmapped areas, as well as from precisely dated map units in the Nazas-Rodeo area (Fig. 4C). The distribution area for ignimbrites of this age interval may extend for a significant distance along Highway 40 to the west of Durango beneath the extensive cover of the younger ignimbrites that occupy the central plateau of the Sierra Madre Occidental volcanic field there. The peralkaline rocks mapped to the west and north of Chihuahua City form a tight cluster of ages from 29.6 to 30.0 Ma.

Finally, along the westernmost 50 km of area 3, 40Ar/39Ar ages of 24 Ma were obtained for four samples from three distinct ignimbrites that cap the Sierra Madre Occidental volcanic field there (Fig. 4C). Near Yécora, at the westernmost end of area 1 (Fig. 4A), there are similar-age rhyolites, dated only by K-Ar at 22.5–24.0 Ma. Together, these appear to represent the last gasp of ignimbrite volcanism within the Sierra Madre Occidental volcanic field.

In Figure 6, the relative probability plots for the three areas are arranged from north to south and adjusted to a common time axis. For each area, a pattern of peaks and gaps in the data is evident. Although some peaks may be exaggerated by concentrated sampling, and some gaps may narrow or disappear upon further sampling and dating, a pattern of waxing and waning volcanic activity nevertheless appears to be a dominant characteristic of the Sierra Madre Occidental volcanic field. From area to area, these fluctuations are not entirely coincident. For area 2, this is to be expected from its north-south orientation. However, areas 1 and 3 both provide a view of cross-arc trends. Combined, they indicate that the time interval of most intense activity was between 36 and 30 Ma, a period that is mostly barren of ages from within area 2. A significant hiatus between 28 and 24 Ma is also common to all three areas. Because the sampling is based on the mapping and has included most of the exposed ignimbrites in individual areas, the quiet intervals may be the more significant and enduring feature of these trends.

The patterns in Figure 6 clearly show that the brief and intense bursts of volcanic activity in the Sierra Madre Occidental volcanic field were separated by relatively quiet periods that were apparently of equal or greater length. This pattern is regionally significant, although in detail the pulses do not seem to be everywhere coincident. The pulses are recorded in generally conformable sequences of ignimbrites that are oriented in belts parallel to the long axis of the volcanic field as a whole. The oldest of these belts is 45–40 Ma and is prominent along the eastern margin of the field, although its western extent is largely concealed. The belt of most intense activity, between 36 and 30 Ma, was distributed more widely across the field and dominates the axial part of the Sierra Madre Occidental volcanic field. This interval is characterized by two major bursts of activity, centered in time at ca. 33 and 30 Ma. Further studies may result in merger of these two into a regional “megapulse.” The youngest belt of volcanism (ca. 24 Ma) was barely encountered in area 1, though it was reached more extensively in mapping and sampling within area 3. It clearly lies along the western half of the Sierra Madre Occidental volcanic field.

In Figure 7, the cross-arc distributions of ages from areas 1 and 3 are shown at the same scale. The location of the known source for the Vista tuff is shown with a blue asterisk in area 3. In this perspective, the ages from area 2 would project onto a narrow belt at the eastern margin of the two profiles. A general westward progression of the central axis of the volcanic arc is apparent and appears to have occurred at a similar rate in each area. The combined impression from Figures 6 and 7 is of three belts “shingled” into a westward pattern and interrupted by periods of relatively sparse volcanism.

A More Regional Perspective

The age trends for the central and northern Sierra Madre Occidental volcanic field (this study) are compared to those published from other areas of the ignimbrite flare-up in Figure 8 (see Table 4 for age summaries and references). Ferrari et al. (1999, 2002, 2007) compiled numerous K-Ar and 40Ar/39Ar ages from the southern part of the Sierra Madre Occidental volcanic field. Their data are based primarily upon reconnaissance sampling within a southwest-trending transect. The dominant period of activity was early Miocene (20–24 Ma) across much of the arc. The youngest of these ignimbrites (22–20 Ma) are prominently exposed adjacent to the Gulf of California. Tuffs of this age have been mapped and dated in once-adjacent areas of southern Baja California (Hausback, 1984; Umhoefer et al., 2001). Ages older than 30 Ma were limited, perhaps due to cover, and were found in samples only from the eastern portion of this transect. Volcanic rocks of Eocene age are known only to the northeast, in the Mesa Central area of central Mexico (Ferrari et al., 1999). The trends of magmatism for the southern Sierra Madre Occidental volcanic field are similar to those noted in this paper, but the overlap of younger belts of volcanism appears to be somewhat greater. The existence of discrete time gaps is not obvious in the data set.

The trend of the Sierra Madre Occidental volcanic field toward the north extends into the Boot Heel and Datil-Mogollon volcanic fields of southwestern New Mexico and southeast Arizona (Fig. 1, inset). Numerous 40Ar/39Ar ages are available for these fields (McIntosh et al., 1992; McIntosh and Bryan, 2000; Chapin et al., 2004). The earliest interval of activity (46–42 Ma) found in the Sierra Madre Occidental volcanic field does not appear to be present in these fields, but the later periods of activity and quiescence correspond well in both regions (Fig. 8; Table 4).

The Trans-Pecos Texas volcanic field, which also includes volcanism adjacent to the Rio Grande in eastern Chihuahua State, lies to the east of the northern Sierra Madre Occidental volcanic field. Earliest ignimbrite volcanism appeared there at ca. 43 Ma (Henry and McDowell, 1986). Starting at 38 Ma, volcanism became more voluminous, with active intervals from 38 to 35 Ma, from 34 to 32.5 Ma, and 31 to 28 Ma (Henry and McDowell, 1986; Miggins, 2009). No major ignimbrites younger than 28 Ma are known. A general trend from NE to SW has been noted in the Trans-Pecos field (Henry et al., 2010). This direction of “younging” is the same as that in the Sierra Madre Occidental volcanic field, but the trend appears to break and restart at 45 Ma near Chihuahua City, along the eastern margin of the Sierra Madre Occidental volcanic field.

Other areas of the ignimbrite flare-up show similar though not coincident patterns of activity (Fig. 8). In the San Juan–Southern Rocky Mountain field of Colorado, volcanism began at 37 Ma, with distinct peaks at 34–32, 29–25, and 22 Ma (Lipman, 2007). In the Great Basin of Utah and Nevada, activity began at ca. 45 Ma, and peaks occurred at 42–40, 35, 31.5–23.5, and 19–15 Ma (Best and Christiansen, 1991; Henry, 2008). Overall, these regions of the ignimbrite flare-up display very similar patterns of intense activity separated by lulls of 1–5 m.y. duration. These peaks and pauses are not fully coincident, but the patterns are remarkably similar.

Over several decades, many attempts have been made to link Late Cretaceous–Tertiary (i.e., Laramide) and Middle Tertiary magmatism in western North America to the subduction history along the Farallon–North American plate boundary (e.g., Coney and Reynolds, 1977; Keith, 1978; Damon et al., 1981, 1983; Humphries, 2009). Specifically, the westward sweep of magmatism indicated by exposures of the Middle Tertiary ignimbrite flare-up in the western United States (e.g., Henry et al., 2010) has been linked to patterns of delamination and/or rollback of the subducted Farallon slab. In a general way, the history observed for the Sierra Madre Occidental volcanic field also reflects this pattern, although the limited recognition of caldera sources for major ignimbrite eruptions in the Sierra Madre Occidental volcanic field precludes a more precise comparison. A more challenging and significant problem is to understand the alternating pattern of brief episodes of intense and quiescent volcanism throughout the area of the Tertiary ignimbrite flare-up in a regional tectonic context.

CONCLUSIONS

Restricting 40Ar/39Ar geochronology to mapped areas helps to assure that each data set documents a complete stratigraphy. Although this compromise limits the extent to which our observations can be applied to the entire Sierra Madre Occidental volcanic field, the limited amount of appropriate 40Ar/39Ar dating available from outside of the three areas that we have described lacks stratigraphic context. Therefore, this integrated study allows the first assessment of two important factors in the evolution of the northern and central Sierra Madre Occidental volcanic field: changes in the intensity of volcanic activity with time, and migration of magma sources through time.

Precise 40Ar/39Ar dating of sanidine and anorthoclase phenocrysts from ignimbrites sampled within three detailed mapping transects within the central and northern Sierra Madre Occidental volcanic field reveals a pattern of short periods of activity separated by quiescent intervals of apparently equal or longer duration. Three, or possibly four, major pulses have been identified: from 46 to 40 Ma, along the eastern margin of the Sierra Madre Occidental volcanic field; from 24 to 22 Ma along the western margin; and the two large pulses closely spaced in time between 36 and 28 Ma that may have blanketed the entire width of the Sierra Madre Occidental volcanic field. These patterns reflect a waxing and waning of volcanism that may be characteristic of the entire Sierra Madre Occidental volcanic field.

Because so few links between calderas and important eruptive units have been documented in the Sierra Madre Occidental volcanic field, the patterns of source migration there can be examined only in general terms. The belts of intense activity overlap significantly in space but appear to show a “shingling” of successively younger belts toward the west. This general trend of east to west migration of discrete ignimbrite belts is compatible with that noted for the ignimbrite flare-up in the western United States (Henry et al., 2010). Such trends have been related to behavior of the subducted Farallon slab beneath western North America.

The episodic behavior for volcanism in the Sierra Madre Occidental volcanic field is also apparent in the rest of the ignimbrite flare-up province in the western United States and may be typical of large continental ignimbrite provinces. This feature may prove to be more challenging to understand in the context of subducted slab behavior and magmatic arc evolution in continental convergent-margin settings.

The sudden and brief appearance of peralkaline ignimbritic volcanism at ca. 30 Ma in a restricted area of central Chihuahua State is an anomaly with respect to both timing and location. Further examination of the extent and age of these rocks, and their relationship to structural features, may establish a separate tectonic context than that for the calc-alkaline volcanism in the remainder of the Sierra Madre Occidental volcanic field.

Knowledge of the full history and caldera distribution for the Sierra Madre Occidental volcanic field remains limited in comparison to its sister fields in the United States. Further advances depend on additional mapping projects that establish the location of caldera sources for eruptive units, and the documentation of complete stratigraphic sections as a context for precise (i.e., ±0.0X Ma) geochronology. The limiting factor for future progress is the continuation of detailed mapping in the Sierra Madre Occidental volcanic field.

The analyses presented this paper are based upon field relations from 22 graduate student projects at The University of Texas at Austin, as well as from independent mapping by Richard Mauger and students of East Carolina University. The University of Texas at mapping was supported by a series of research grants from the National Aeronautics and Space Administration, the National Science Foundation, and the Geology Foundation, Department of Geological Sciences of the University of Texas at Austin. The 40Ar/39Ar age determinations were carried out at the New Mexico Geochronology Laboratory with the help of Richard Esser and Lisa Peters. The age determinations were funded in part by a research grant from the ExxonMobil Corporation and funding from the Jackson School of Geosciences of the University of Texas at Austin. Christopher Henry, Richard Mauger, and Alexander Iriondo provided valuable reviews of an earlier version of this manuscript. Final reviews by Henry and Daniel Miggins further improved the clarity of the paper.

1Supplemental Table 1. Excel file of revision of previously published 40Ar/39Ar analytical data. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00792.S1 or the full-text article on www.gsapubs.org to view Supplemental Table 1.
2Supplemental Table 2. Excel file of new 40Ar/39Ar analytical data. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00792.S2 or the full-text article on www.gsapubs.org to view Supplemental Table 2.
3Supplemental Figure 1. Age-probability distribution spectra, % radiogenic 40Ar contents, 39Ar concentrations, and K/Ca ratios are given for published analyses for which new measurements have been added, or which have been renormalized to 28.02 Ma for the Fish Canyon sanidine standard since publication. Open symbols denote analyses not included in the age calculation. The weighted mean ages are shown with 2σ error limits. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00792.S3 or the full-text article on www.gsapubs.org to view Supplemental Figure 1.
4Supplemental Figure 2. The same information as in Supplemental Figure 1 for 40Ar/39Ar ages published here for the first time. For published ages that have remained unchanged since publication, the equivalent data are given in the original publications. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00792.S4 or the full-text article on www.gsapubs.org to view Supplemental Figure 2.