The White River Ash, a prominent stratigraphic marker bed in Alaska (USA) and Yukon (Canada), consists of multiple compositional units belonging to two geochemical groups. The compositional units are characterized using multiple criteria, with combined glass and ilmenite compositions being the best discriminators. Two compositional units compose the northern group (WRA-Na and WRA-Nb), and two units are present in the eastern group (WRA-Ea and the younger, WRA-Eb). In the proximal area, the ca. 1900 yr B.P. (Lerbekmo et al., 1975) WRA-Na displays reverse zoning in the glass phase and systematic changes in ilmenite composition and estimated oxygen fugacity from the base to the top of the unit. The eruption probably tapped different magma batches or bodies within the magma reservoir with limited mixing or mingling between them. The 1147 cal yr B.P. (calibrated years, approximately equivalent to calendric years) (Clague et al., 1995) WRA-Ea eruption is only weakly zoned, but pumices with different glass compositions are present, along with gray and white intermingled glass in individual pumice clasts, indicating the presence of multiple magmatic bodies or layers. All White River Ash products are high-silica adakites and are sourced from the Mount Churchill magmatic system.


The White River Ash (WRA) is an important Holocene chronostratigraphic marker throughout eastern Alaska (USA) and in Yukon and western Northwest Territories (Canada) (Lerbekmo and Campbell, 1969; Péwé, 1975), and provides age control for peat studies (Harris and Schmidt, 1994; Robinson and Moore, 1999, 2000), glacial fluctuations (Denton and Karlen, 1977), paleoecological studies (Slater, 1985), and archaeological and anthropological studies (Workman, 1979; Moodie et al., 1992) (Fig. 1). WRA has been defined by its geographic distribution and its stratigraphic position, combined with radiocarbon dating, with less emphasis on its petrographic and geochemical features. The northern lobe erupted between A.D. 150 and 500 (Jensen and Froese, 2006) or ca. 1900 yr B.P. (Lerbekmo et al., 1975), while the eastern lobe erupted A.D. 803 or 1147 cal. yr B.P. (Clague et al., 1995). Cryptotephra studies in lake and peat cores have extended the distribution of the eastern lobe to southeastern Alaska (sites A–C, Fig. 1), northern British Columbia (site D, Fig. 1), Newfoundland, northern Europe, and Greenland (Lakeman et al., 2008; Payne et al., 2008; Addison et al., 2010; Pyne-O’Donnell et al., 2012; Jensen et al., 2012). In addition, cores in southeastern Alaska contain two other tephra beds with compositions similar to WRA. These are the ca. 300 yr B.P. (ca. A.D. 1650) Lena ash preserved at sites A and B (Fig. 1) and the ca. 6330 cal yr B.P. MTR-146 ash preserved at site B (Fig. 1) (Payne et al., 2008). The distribution of the WRA indicates that its volcanic source is in the Wrangell volcanic field close to the Yukon-Alaska border (Fig. 1).

Unfortunately, glass compositions are similar in the northern and eastern lobes of the WRA tephra, and published analyses cover a significant range, making it difficult to assign a particular tephra sample to a specific lobe (Addison et al., 2010; Payne et al., 2008; Lakeman et al., 2008; Froese and Jensen, 2005; Richter et al., 1995; Begét et al., 1992; Downes, 1985). Downes (1985) could not find any systematic variations with stratigraphic position in the glass composition in either the northern or eastern lobes, and attributed the wide range of values within each to inhomogeneity in the magma chamber. Northern and eastern lobe WRA tephra can be distinguished through the compositions of the ilmenite phenocrysts (Richter et al., 1995; Lerbekmo et al., 1975). Lerbekmo et al. (1975), using Fe-Ti oxide geothermometry, noted systematic variations in preeruptive temperature with stratigraphic position at three northern lobe locations. Few published analyses exist of both glass and Fe-Ti oxide phases, making it difficult to evaluate the compositional heterogeneity of the WRA. Only a small number of northern lobe samples have glass compositional data (Froese and Jensen, 2005; Downes, 1985).

The exact vent area of the WRA has been disputed, and two possible locations have been suggested. Lerbekmo and Campbell (1969) postulated that the WRA vent was located on the floor of a deep valley beneath the Klutlan Glacier near what they identified as a large pumice mound (4 in Fig. 2). McGimsey et al. (1992) and Richter et al. (1995) questioned the likelihood of a vent located at the floor of a valley lacking an edifice or nearby volcano, and proposed that Mount Churchill was the vent. Lerbekmo (2008) questioned whether Mount Churchill was a volcano, and reiterated that the vent for WRA was underneath the Klutlan Glacier.

We document the petrography, compositional range of the glass and Fe-Ti oxide phases, and Fe-Ti oxide geothermometry in a suite of WRA samples to determine whether systematic chemical heterogeneity and/or variation in temperature and oxygen fugacity are present. New Nd and Pb isotopic analyses for selected WRA samples and for the dacite lava flow at the summit of Mount Churchill are compared to volcanic products younger than 5 Ma in the Wrangell volcanic field to try and determine magmatic sources. We reevaluate the pumice and dacite flow at the summit of Mount Churchill and their implications for the source of the WRA and whether Mount Churchill is a volcano.


Volcanoes in the Wrangell volcanic field range in age from 26 Ma at the southeastern end near the Yukon border to ca. 0.2 Ma at the northwestern end, and are related to subduction of the Pacific plate and Yakutat terrane beneath North America (Richter et al., 1990). The Wrangell volcanic field is part of the larger alkaline to calc-alkaline Wrangell volcanic belt that extends from Alaska through the Yukon and into northern British Columbia (Skulski and Francis, 1991). The majority of Wrangell volcanic field volcanism ceased ca. 0.1–0.2 Ma when most of the motion between the Pacific plate and North American plate became accommodated through strike-slip faulting and thrusting (Richter et al., 1990). Volcanic activity at Mount Churchill, located in the southeast Wrangell volcanic field, and the widespread WRA are exceptions to the southeast to northwest age progression in the Wrangell volcanic field (Fig. 2).

Extrusive igneous materials in the Wrangell volcanic field are divided into three separate geochemical suites (Preece and Hart, 2004). Trend 1 is a transitional tholeiitic suite with high TiO2, Y, and Zr contents that is restricted to areas of localized intra-arc extension at Mount Sanford and the interior mesas area (Figs. 2 and 3). Trend 2a is a calc-alkaline suite found throughout the Wrangell volcanic field with low TiO2, Y, and Zr contents. Trend 2b consists of a suite of calc-alkaline, high-silica adakitic and transitional dacitic materials with elevated La/Yb and Sr/Y ratios, and elevated Mg# and Ni contents restricted to Mount Drum, the westernmost vent in the Wrangell volcanic field, and Mount Churchill (Figs. 2 and 3). As discussed in Preece and Hart (2004), the high-silica adakitic samples fulfill the criteria for high-silica adakite as defined by Defant and Drummond (1990) and Martin et al. (2005). WRA also belongs to trend 2b (Preece and Hart, 2004; Westgate et al., 2008; see data in Table S1 in the Supplemental File1). At Mount Drum, high-silica adakite production results from the formation of a slab window elevating temperature gradients at the edge of the subducting plate (Preece and Hart, 2004; Eberhart-Phillips et al., 2006). At Mount Churchill, high-silica adakite production is probably linked to stalled subduction, which allowed temperature gradients in the subducting plate to increase until melting occurred (Preece and Hart, 2004).


The terms northern lobe and eastern lobe refer to the geographic distribution of WRA ash. In this contribution, the terms WRA-N (northern) and WRA-E (eastern) are used to designate the distinct compositional groups within the WRA, and distinguish them from the geographic distribution of WRA tephra-fall deposits (lobes). WRA-N includes two geochemical units (WRA-Na and WRA-Nb) as does WRA-E (WRA-Ea and WRA-Eb). The geochemically defined groups do not imply stratigraphic positions within the WRA deposits unless specifically mentioned. Many of the samples used in this study were previously reported by Lerbekmo et al. (1975), Denton and Karlen (1977), or Richter et al. (1995); 14 new samples are included. Table 1 lists the locations for all samples used in this study. The location number in Table 1 corresponds to the number in Figure 2 and to the sample number. Samples from different stratigraphic horizons at the same location are indicated by capital letters with lettering starting at the base of the section. For example, 9A and 9B refer to two different samples from location 9. Lower case letters refer to analyses of individual pumice clasts from a given location. For example, 4a identifies a single pumice clast from location 4.

Exposures at the summit of Mount Churchill are ephemeral and highly dependent on snow accumulation and wind scouring (see Table 1 and Fig. 2 for the locations). The summit of Mount Churchill forms the western wall of a 4.2 by 2.7 km basin (Fig. 4A). Along the rim of the basin is an intermittently exposed 119 ± 17 ka in situ dacite lava flow (Figs. 4A, 4B; Richter et al., 1995). Pumice and lithics drape the base of the dacite outcrop. Thick deposits of pumice and lithic blocks, individually to at least 50 cm, are frozen in a matrix of tephra on the rim of the basin and in adjacent areas (Figs. 4C–4E). The lithic bombs and blocks consist of Wrangell volcanic field–type volcanic rocks and granodiorites similar to rocks mapped as underlying the massif (MacKevett, 1978). On the west wall of the basin below the summit of Mount Churchill are three outcrops of highly altered, possibly columnar jointed, nonstratified rock that is consistent with altered lithic clasts observed in the proximal WRA deposits (Fig. 4F). Approximately 1 km northeast of the basin, the exposed bedrock is steeply east dipping, columnar jointed, thick lava flows (Figs. 4G, 4H), unlike the horizontal to gently dipping, nearby older rocks of the Wrangell volcanic field.

At all Klutlan Glacier locations there are two types of pumice clasts (see Table 1 for stratigraphic details; see Fig. 2 for locations). The light gray pumice clasts typically are less vesicular than the white pumice clasts. At location 6, one pumice clast is composed of an admixture of white and gray glass, suggesting magma mingling (Fig. 5).

Location 4 is the eastern lobe pumice mound of Lerbekmo and Campbell (1969) adjacent to their proposed vent area (4 in Fig. 2). The pumice mound is a 30–40 m tephra deposit sitting on a bedrock bench (Fig. 6). No stratification or other primary air-fall sedimentary structures were observed at this site.

Primary air-fall stratification is observed at location 7, where multiple samples were collected from a 4.8-m-thick, bedded deposit consisting of a series of fining-upward beds (7 in Fig. 2; Fig. 7). Figure 7E shows the open-framework, clast-supported structure present in the coarser parts of the deposit. The deposit is thickly covered by >50 m of reworked WRA that sloughed off the high, steep, adjacent valley walls, and may also include reworked pumice deposits from other processes (Fig. 7A; cf. Donaldson et al., 1996).

The WRA consists of a single light colored tephra layer in most distal exposures, but in a few places both the northern and eastern lobes are preserved in stratigraphic succession. One important section occurs at location 12 (Fig. 2) (Hughes et al., 1972; Lerbekmo et al., 1975). Three tephra beds are present in the section (Fig. 8), the northern lobe at the bottom, the eastern lobe in the middle, and the uppermost tephra bed. Lerbekmo et al. (1975) interpreted both the upper and middle tephra beds as the eastern lobe, but the presence of 10 cm of peat between the middle and upper tephra beds suggests that these are separate eruptive deposits separated by sufficient time for appreciable sediment accumulation.



Tephra samples of the WRA contain phenocrysts of plagioclase, amphibole, magnetite, ilmenite, and trace amounts of orthopyroxene and apatite within highly vesicular, frothy colorless glass (Table S8 in the Supplemental File [see footnote 1]). Biotite occurs in the eastern lobe in trace amounts (<∼1%) in some pumice clasts from locations 4 and 6 (Fig. 2). Biotite was not found in any of the northern lobe samples, although two samples contained trace amounts of reddish-brown phenocrysts tentatively identified as oxyhornblende (Table S8 in the Supplemental File [see footnote 1]). Biotite is abundant in pumice samples from samples 2 and 3 from the summit of Mount Churchill and is most common in pumice clasts from location 8 (Fig. 2). Trace amounts of biotite occur in sample 12E, the uppermost tephra bed at location 12 (Fig. 2).

Major Element Composition of Fe-Ti Oxide Minerals and Glasses

Previous studies have demonstrated that the eastern and northern lobes of the WRA have different ilmenite compositional ranges (Lerbekmo et al., 1975; Downes, 1985; Richter et al., 1995) and this distinction has been used to define geochemical groups WRA-E and WRA-N. The most comprehensive set of ilmenite analyses for the WRA is from Lerbekmo et al. (1975). Compositions reported by Lerbekmo et al. (1975) are averages of ∼10 grains per sample. Averages tend to show a more limited compositional range than individual analyses. In Figure 9 individual analyses (Table S4 in the Supplemental File [see footnote 1]) for previously identified samples (samples used by Lerbekmo et al., 1975; Richter et al., 1995) are plotted against fields defined by the ilmenite compositions from Lerbekmo et al. (1975). Most of the WRA-E samples plot within the field defined by the Lerbekmo et al. (1975) data, but samples from location 4 (sample 4) have a slightly wider range (Fig. 9A; Table S4 in the Supplemental File [see footnote 1]). The WRA-N samples in Figure 9B show a significantly larger range in TiO2 contents compared to the field defined by the Lerbekmo et al. (1975) data. Figure 9 suggests that caution must be exercised identifying WRA samples when ilmenite compositions are in the 31–32 wt% TiO2 range.

Ilmenite compositions were used to classify the new samples into WRA-N or WRA-E groups (Fig. 10). Both of the summit samples (samples 2 and 3) belong to WRA-E (Fig. 10A) as do samples from location 7 (samples 7A, 7E, and 7G) and location 8 (sample 8) (Fig. 10B). Ilmenite in sample 8 has some of the lowest TiO2 contents of any of the studied samples (Fig. 10B; Table 2; Table S4 in the Supplemental File [see footnote 1]). At location 12 both the lowermost WRA-N tephra bed (sample 12A), and the middle WRA-E bed (samples 12B, 12C, and 12D) plot as expected (Fig. 10C). The uppermost tephra bed (sample 12E) overlaps the compositional range of WRA-E, but is skewed to lower TiO2 contents. The difference in the distribution of the analytical points indicates that the uppermost bed (sample 12E) is not reworked tephra from WRA-E. Compared to all other WRA-E samples, both the uppermost bed at location 12 (sample 12E) and sample 8 (8 in Fig. 2) have similar low TiO2 contents. In addition, sample 12E is 10 cm above WRA-E at location 12. For these reasons, sample 12E and sample 8 are classified as geochemical group WRA-Eb, while the remaining WRA-E samples are grouped as geochemical group WRA-Ea. Data for the distal samples from this study are presented in Figure 10D; all belong to WRA-N.

Titanomagnetite compositions differ little among the samples (Table 3; Table S3 in the Supplemental File [see footnote 1]). Titanomagnetite compositions in WRA-N have TiO2 contents between 5 and 7 wt% and in WRA-Ea and WRA-Eb have a slightly wider range of TiO2 contents, from 3 to 6.5 wt%.

The compositions of glass shards from WRA-N, WRA-Ea, and WRA-Eb display a wide range of values (Table 4; Table S2 in the Supplemental File [see footnote 1]). In WRA-N, WRA-Ea, and WRA-Eb individual samples usually have a restricted compositional range, but when displayed on scatter plots the distributions overlap, producing a continuous array of values (Fig. 11). The proximal and distal samples of WRA-N have similar ranges of SiO2 from ∼71–78 wt%, but several samples show a compositional gap between 73.5 and 75.9 wt% (Fig. 11C). Sample 16 has a much more restricted range of SiO2 contents with only 2 analyses <75.5 wt% (Fig. 11D). All of the proximal and distal WRA-N samples, except sample 16, are referred to as geochemical group WRA-Na. WRA-Nb consists of sample 16.

A slightly more restricted range from ∼72.5 to 76.5 wt% SiO2 is observed for glass from WRA-Ea (Fig. 11E; Table 4). WRA-Ea sample 3 from the summit of Mount Churchill extends to higher silica contents compared to other WRA-Ea samples (Fig. 11E; Table 2). WRA-Eb samples 12E and 8 have slightly different silica distributions (Fig. 11F).

Fe-Ti Oxide Geothermometry

Magmatic temperature and oxygen fugacity estimates for the WRA samples are plotted in Figure 12 (individual estimates are listed in Table S5 in the Supplemental File [see footnote 1]. WRA-N and WRA-E samples form arrays that proceed in different directions from a region of overlap at ∼820 °C and ∼1.47 ΔNNO. Within WRA-E, decreasing temperatures (∼820–700 °C) are associated with a slight increase in oxygen fugacity (∼1.46–1.59 ΔNNO) with the most silicic samples, WRA-Eb, having the lowest temperature estimates. The WRA-N samples show different behavior. There is a substantially larger decrease in oxygen fugacity (∼1.48–0.9 ΔNNO) with a smaller decrease in temperature (∼850–750 °C). One outlier exists within the WRA-N samples (N. outlier arrow in Fig. 12). In general, within the proximal WRA-Na samples, both temperature and oxygen fugacity decreased with increasing average silica content of the melt.

Trace Element Composition of Glass

Previous studies employing trace element analyses of bulk glass concentrates have found little difference between eastern and northern lobe samples (Westgate et al., 2008). Individual and average trace element contents for samples in this study are extremely similar and most glass is high-silica adakite (Table 5; Fig. 13A; Table S6 in the Supplemental File [see footnote 1]). WRA-Ea sample 4 and WRA-Nb sample 16 have slightly lower average La/Yb and plot just outside the high-silica adakite range. Within WRA-N, average La/Yb is restricted between 19 and 23 and tends to decrease with increasing SiO2 content (e.g., for location 11, compare 11A at the base, 72.4 wt%, to 11C at the top, 76.3 wt% (Table 5). WRA-E samples cover a similar range in average Sr/Y ratios, but the average La/Yb range is wider, from 19 to 34, and there is no correlation with average SiO2.

WRA-N and WRA-E have similar patterns and concentrations on averaged rare earth element (REE) plots (Figs. 13B–13D). All samples display steep slopes from La to Dy or Ho and are either flat or have a slight increase in chondrite-normalized concentrations between Er and Yb or Lu. Slightly positive to slightly negative Eu anomalies are present in both WRA-E and WRA-N.

Trace element contents are very similar between the WRA-N and WRA-E, but some differences are observed in Zr and Th behavior. With increasing SiO2 contents, WRA-N Zr and Th concentrations maintain a constant range (Figs. 14A, 14C), while WRA-E samples display decreasing Zr and increasing Th contents (Figs. 14B, 14D).

Isotopic Analysis

As part of a regional study, selected samples from vents younger than 5 Ma in the Wrangell volcanic field were analyzed for Sr, Nd, and Pb isotope ratios (Preece and Hart, 2004; Westgate et al., 2008; Table 6; Table S7 in the Supplemental File [see footnote 1]). Other than Mounts Churchill and Drum, the only known source of high-silica adakite tephra in central or eastern Alaska is Hayes Volcano, the most northern vent in the Alaska Peninsula (Preece and Hart, 2004; Westgate et al., 2008); therefore, isotope ratios from McHugh et al. (2012) for Hayes Volcano are presented. The Mount Churchill dacite lava flow (1 in Fig. 2), and WRA-Ea samples 4b (4 in Fig. 2) and UA459 (from Holmes Creek; see Lerbekmo et al., 1975, for sample details) are offset to higher 87Sr/86Sr and lower εNd compared to most other Wrangell volcanic field igneous materials. The only sample with ratios similar to WRA-Ea and the Mount Churchill dacite flow is a pumice sample from the Skookum Creek volcanic complex (93SJP2) (Fig. 15A). Tephra, dacite, and rhyolite from Hayes Volcano in the eastern Aleutian Arc have 87Sr/86Sr and εNd values similar to those of Mount Churchill and the WRA (Westgate et al., 2008; McHugh et al., 2012; Preece and Hart, 1991, personal data). In Figures 15B and 15C, the Mount Churchill dacite lava flow and WRA-Ea are offset to higher 206Pb/204Pb compared to other vents in the Wrangell volcanic field, including the Skookum Creek volcanic complex pumice and high-silica adakite materials from Mount Drum. They are also offset from Hayes Volcano. This indicates that Sr, Nd, and Pb isotope ratios can be used to help identify the WRA and can discriminate between the high-silica adakite tephra producing vents at Mount Churchill, Mount Drum, and Hayes Volcano.


During an eruption or eruptive phase, magmatic composition and/or parameters can change. This can be examined if samples are taken at different stratigraphic levels through a primary tephra bed or set of tephra beds.

Samples from WRA-Na

The greatest chemical heterogeneity in WRA-Na is at location 11 (Fig. 2), where the average SiO2 content of the glass increases from 72.4 wt% at the base (sample 11A) to 76.3 wt% at the top of the bed (sample 11C) (Table 4; Fig. 16A; Table S2 in the Supplemental File [see footnote 1]). The compositions of the basal (11A) and middle (11B) samples overlap. There is an ∼1 wt% SiO2 gap between the middle (11B) and top (11C) samples, or a 3 wt% gap in average SiO2 contents (Table 4). This compositional gap may be an artifact of the sampling procedure. From the base to the top of the deposit Sr contents decrease (Fig. 16B; Table 5; Table S6 in the Supplemental File [see footnote 1]), Y, Zr, Pb, and most middle and heavy Rare Earth Elements (REEs) show no change (Fig. 16C; Table 5), and Rb, Ba, Cs, U, Th, Hf, Ta, and light REE contents increase very slightly (Fig. 16D; Table 5). The TiO2 contents of ilmenite increase ∼4.4 wt% from the base to the top (Table 2, Fig. 16E; Table S4 in the Supplemental File [see footnote 1]). Compared to the basal sample (11A), a wider range of titanomagnetite FeOt (total Fe as FeO) contents is present in the middle (11B) and top (11C) samples, with the highest FeOt contents in the top sample (Fig. 16F; Table 3; Table S3 in the Supplemental File [see footnote 1]). Estimates of the temperature and oxygen fugacity are listed in Table S5 in the Supplemental File (see footnote 1) and plotted in Figure 16G. The basal sample (11A) forms a short linear array defining the highest oxygen fugacity at a given temperature, while the middle (11B) and top (11C) samples display more scatter. Estimates from the middle sample (11B) are between the linear arrays defined by the basal and top samples.

A similar pattern occurs at location 10 near the terminus of the Russell Glacier, where the base of WRA-Na (sample 10A) has slightly lower SiO2 contents in glass shards than the top of the bed (sample 10B) (Table 4; Fig. 16H; Table S2 in the Supplemental File [see footnote 1]). The SiO2 range at this location is nearly the same as that of the basal and middle samples from location 11. Ilmenite grains from the base of the bed (10A) have restricted TiO2 contents, from 33.8 to 36.0 wt%, compared to the significantly wider range in the upper sample (10B), from 32.8 to 40.2 wt% (Table 2; Table S4 in the Supplemental File [see footnote 1]). Titanomagnetite compositions from the base of the bed (10A) are offset to slightly lower FeOt contents compared to the upper sample (10B) (Table 3; Fig. 16I; Table S3 in the Supplemental File [see footnote 1]). Estimates from the base of the bed (10A) form a short linear array, while estimates from the top of the bed (10B) are more scattered, and most have slightly lower oxygen fugacity and temperature estimates (Table S5 in the Supplemental File [see footnote 1]; Fig. 16G).

In contrast to locations 11 and 10, there is no geochemical variation in WRA-Na between the base (sample 9A) and top (sample 9B) at location 9 (Fig. 2; Tables 2–4; Tables S2–S4 in the Supplemental File [see footnote 1]). The silica contents of these two samples are most similar to basal samples from locations 10 and 11 (Table 2; cf. 9A and 9B to 10A and 11A). Most likely, only a portion of WRA-Na has been preserved at location 9.

WRA-Ea and WRA-Eb

At location 7 (Figs. 2 and 7B), three stratigraphic horizons of WRA-Ea (samples 7A, 7E, and 7G) were analyzed and no variations in glass, ilmenite, titanomagnetite, estimated temperature, or estimated oxygen fugacity were found (Tables 2–4; Tables S2–S5 in the Supplemental File [see footnote 1]). In contrast, at location 12 (Fig. 2), WRA-E consists of two discrete tephra beds, WRA-Ea and WRA-Eb, separated by 10 cm of peat (Fig. 8). Within WRA-Ea the maximum SiO2 concentration of individual glass shards increases slightly with height in the deposit, with significantly higher concentrations in WRA-Eb (Fig. 17A). Stratigraphic compositional zonation is present in glass trace elements, with minimum Sr and Y concentrations decreasing, and maximum Th concentrations increasing, from the base of WRA-Ea to WRA-Eb (Fig. 17). For WRA-Ea, estimated temperatures for the top sample (12D) are slightly lower than the middle (12C) and basal (12B) samples, suggesting tapping of somewhat cooler magmas as the eruption progressed (Table S5 in the Supplemental File [see footnote 1]; Fig.17I). Magmatic temperatures for WRA-Eb (sample 12E) extend to cooler values still, attaining the lowest temperatures at this location (Fig. 17I).


Summary of WRA Tephra Beds

Multiple geochemical criteria need to be used to distinguish the different geochemical groups in WRA. Ilmenite compositions clearly separate WRA-N from WRA-E (Figs. 9 and 10). WRA-N glass has slightly lower SiO2 contents than that of WRA-E and there are minor differences in the trace element contents (Figs. 11, 13, and 14). The differences in the glass compositions are subtle and cannot be used to consistently separate WRA-N from WRA-E. Glass and Fe-Ti oxide composition and, in some cases, temperature and oxygen fugacity estimates separate the geochemical units within WRA-N or WRA-E (Table 7).

WRA-Na corresponds to the ca. 1900 yr B.P. (Lerbekmo et al., 1975) northern lobe. WRA-Na has a wide compositional range resulting from increasing silica content of the melt over the duration of the eruption (see data for site 11). At any one location a different portion of the chemical range may be found depending on preservation and sampling. Plume heights, wind direction, and later reworking of the tephra may result in only a portion of the eruption being preserved at a given location.

In the distal Sixtymile area (locations 13–22 in Fig. 2) two geochemically distinct units of WRA-N are present. The relative stratigraphy of the two units is not clear, but given the small geographic area they are from, and their similar geomorphic setting, their depth below the surface may be used to indicate a possible age relationship. WRA-Na samples are typically ∼50 cm below the surface. WRA-Nb (sample 16) is 20–30 cm below the surface, suggesting it preserves either the last, most silicic, stages of the ca. 1900 yr B.P. (Lerbekmo et al., 1975) northern lobe eruption or a slightly younger eruption from the same magma chamber as the northern lobe.

WRA-Ea samples belong to the widespread 1147 cal yr B.P. (Clague et al., 1995) eastern lobe. At location 12, WRA-Na, WRA-Ea, and WRA-Eb are in stratigraphic succession. WRA-Ea is stratigraphically below and separated from WRA-Eb by 10 cm of silty peat (Fig. 8). WRA-Eb results from a slightly younger eruption, likely from the same magmatic system as WRA-Ea. WRA-Eb is represented by two samples, 12E and 8. It is unclear whether sample 8 is from the same eruption as sample 12E or if it is from a different eruption of the evolving WRA-E magmatic system.

Pumice samples 2 and 3, from the summit of Mount Churchill, belong to WRA-E. Glass and Fe-Ti oxide analyses of sample 2 are most similar to WRA-Ea, the 1147 cal yr B.P. (Clague et al., 1995) eastern lobe (Tables 2–5; Figs. 9 and 12–14). In contrast, sample 3 is classified as WRA-Ea based on its ilmenite compositions, but glass compositions are more silicic than most of the WRA-Ea and are more similar to WRA-Eb at location 12.

A possible correlative of the WRA-Eb geochemical unit is the Lena ash, an ∼300-yr-old tephra bed thought to be from Mount Churchill (Payne et al., 2008). Comparison shows that glass shards from the Lena ash have slightly higher FeOt and CaO contents and slightly lower Na2O contents then WRA-Eb (cf. data in Payne et al., 2008; Fig. 18). Additional analytical work is needed to confirm a correlation.

Petrologic Aspects of the WRA Magmatic System

A detailed discussion of the petrologic evolution of the WRA magmatic system is beyond the scope of this contribution, but some petrologic processes and trends can be identified. In the immediate vicinity of Mount Churchill there are several outcrops of 11–8 Ma nonadakitic calc-alkaline volcanic products that are similar to trend 2a of the Wrangell volcanic field (Trop et al., 2012; cf. Fig. 3). No igneous products between 8 Ma and 119 ka have been documented in the area. This is ascribed to shifts in plate convergence and subduction of the thicker Yakutat microplate causing volcanism to migrate northwestward (Richter et al., 1990; Trop et al., 2012). Plate kinematic models by Madsen et al. (2006) and Thorkelson et al. (2011) postulated the presence of a large slab window extending from Yukon through most of British Columbia and ending just north of Vancouver. Upwelling asthenosphere through the slab window, located ∼150 km east of Mount Churchill, could provide a heat source to the stalled slab, allowing temperature gradients to increase and induce high-silica adakitic melt generation (cf. Thorkelson and Breitsprecher, 2005). Temperature and oxygen fugacity estimates of the lowest silica WRA-Na and WRA-Ea overlap (Fig. 12) and may approximate the conditions present in the slab at the time of melt extraction. Stratigraphic profiles through pyroclastic deposits provide information on chemical and mineralogic gradients within magmatic bodies (Bachmann and Bergantz, 2008, and references therein). A range of geochemical trends can be present: pyroclastic deposits may be homogeneous, chemically zoned from base to top, or show chemical variations between pumices collected at the same stratigraphic horizon (Bachmann and Bergantz, 2008; Cooper et al., 2012). Typically, chemically zoned pyroclastic deposits grade from more silicic material at the base to less silicic material at the top and are interpreted to represent the systematic emptying of a compositionally zoned magma chamber where more fractionated, less dense melt was erupted first followed by less fractionated, denser and more crystal-rich magma as deeper layers in the magma chamber were tapped (cf. Hildreth, 1981). Even when the pattern is more complex, for example the 1912 Katmai eruption (Turner et al., 2010), the basal material is usually the most silicic (Hildreth, 1981; Bachmann and Bergantz, 2008). WRA-Na displays reverse zoning with SiO2 contents increasing by ∼4 wt% from the base to the top of the deposit. This is the opposite pattern of most zoned pyroclastic deposits and cannot be explained by sequential emptying of a zoned magma chamber. One possibility is that the eruption tapped different magma batches or bodies within the magma reservoir. This might help explain the compositional gaps that are present in WRA-Na.

Unlike the strongly zoned ca. 1900 yr B.P. (Lerbekmo et al., 1975) eruption (WRA-Na), the 1147 cal yr B.P. (Clague et al., 1995) (WRA-Ea) eruption is only weakly zoned with a slight increase in maximum silica contents from the base to the top of the deposit at location 12 (Fig. 17A). There is physical evidence of magma mingling in WRA-Ea samples from location 6 (Fig. 5), but glass analyses from the gray and white portions do not show any compositional differences (Figs. 19A, 19B). In contrast, glass analyses from individual pumice clasts from location 4 cluster on bivariate plots (Figs. 19C, 19D), showing clear evidence for at least small chemical differences within the magma, possibly the result of slightly different batches or bodies within the magma chamber that mixed to varying degrees. This is similar to some eruptions from the Taupo Volcanic Zone, New Zealand, where three to four separate magmatic populations are documented in pumice clasts from the Kidnappers supereruption (Cooper et al., 2012) and Whakamaru Group ignimbrites (Brown et al., 1998).

Glass and ilmenite compositions, temperatures, and oxygen fugacity estimates for the younger WRA-Eb samples coincide with and extend linear trends from the earlier WRA-Ea eruption to cooler and more silicic compositions (Figs.11E, 12, 14, and 17), and may be genetically related to the WRA-Ea eruption.

Is the Pumice Mound a Vent Area and the Volcanic Source of WRA?

The pumice mound of Lerbekmo and Campbell (1969) was originally described as a 150–200-m-high pumice cone with pumice blocks to 60 cm diameter. As documented here, visits to the location failed to yield any clast larger than 30 cm or any evidence of a primary origin (e.g., no internal stratification or grain-size sequencing). In addition, ablation of the Klutlan Glacier and shrinkage of the perennial snow cover reveals that the pumice mound is a bedrock bench overlain by 30–40 m of homogeneous tephra (Figs. 6A, 6B). No evidence exists to support a vent adjacent to the mound.

Mount Churchill is a Volcano

Physical evidence strongly points to Mount Churchill as a volcano; petrographic and geochemical data indicate that it was the vent from which WRA erupted. Thick deposits of pumice and lithic blocks are frozen in a matrix of tephra on the summit of Mount Churchill and are consistent only with a near-vent origin. The lithic blocks are composed of the Wrangell volcanic field and granodiorite bedrock mapped as underlying the massif (cf. MacKevett, 1978). Large, violently explosive Plinian eruptions typically bring up bedrock material that underlies the volcano. High winds acting over 1100 yr have frequently scoured any exposed surfaces, winnowing loosened ash, gradually lowering the surface and concentrating the dense, lithic clasts.

The in situ dacite lava flow on the rim (Fig. 4B) and steeply east dipping columnar jointed thick lava flows (Figs. 4G, 4H) of the Mount Churchill volcanic edifice are unlike the horizontal to gently dipping older rocks of the Wrangell volcanic field. Exposures of highly altered, possibly columnar jointed, nonstratified rock (Fig. 4F) that is megascopically consistent with the altered lithic clasts observed in the proximal WRA deposits provide additional evidence that the summit basin is a vent area.

The WRA and the 119 ka dacite flow, both from the rim of the summit basin of Mount Churchill, are geochemically similar. All of these samples are high-silica adakite, and WRA-Ea and the dacite flow have similar isotopic compositions that are different from all other known high-silica adakitic volcanic vents in central and eastern Alaska. This composition is unlike the surrounding older Wrangell volcanic field volcanic products documented in Trop et al. (2012). This also differs from the older alkaline and calc-alkaline volcanic products of the Wrangell volcanic belt located to the east of Mount Churchill (Skulski and Francis, 1991).


Mount Churchill is interpreted to be the source vent of the WRA. Pumice blocks from the summit of Mount Churchill belong to WRA-Ea. All WRA samples and the 119 ka dacite flow from Mount Churchill have high-silica adakitic geochemistry and are interpreted to be derived from melting of a subducted slab. The probable presence of a slab window immediately east of Mount Churchill, accompanied by upwelling asthenosphere, increased the temperature gradient in the adjacent slab, allowing melting to commence. Stalling of the slab in the area may also elevate temperatures in the subducted slab, as suggested in Preece and Hart (2004).

WRA deposits can be assigned to WRA-E or WRA-N, and are best identified using ilmenite compositions. Within WRA-N there are two distinctive compositions, WRA-Na and WRA-Nb. In the proximal area, ca. 1900 yr B.P. (Lerbekmo et al., 1975) WRA-Na displays increasing silica content accompanied by systematic changes in trace element compositions, Fe-Ti oxide compositions, and temperature-fugacity estimates. WRA-Nb may represent either a restricted phase of the WRA-Na or a separate eruption from the Mount Churchill magmatic system. Within WRA-E, WRA-Ea is clearly older than WRA-Eb. The 1147 cal yr B.P. (Clague et al., 1995) WRA-Ea deposits either do not show or only weakly display systematic changes in glass or Fe-Ti oxide composition with stratigraphic position. On geochemical plots, WRA-Eb samples are on linear extensions toward higher silica content and lower temperature estimates compared to WRA-Ea samples, strongly suggesting a genetic link. WRA-Eb samples represent a younger eruption or eruptions from the evolving Mount Churchill magmatic system.

This contribution is dedicated to the memory of Don Richter of the U.S. Geological Survey, who worked extensively in the Wrangell volcanic field and encouraged two of us (Preece and Hart) to examine the volcanology and petrology of the Wrangell volcanic field. Preece thanks D. Froese of the University of Alberta for support during a portion of this study. This work was supported by funding from the Natural Sciences and Engineering Research Council of Canada to Westgate and from the U.S. National Science Foundation to Hart. We thank Keith Labay of the U.S. Geological Survey for preparation of Figures 1 and 2, and Tom Sisson and two other reviewers for their comments.


Large pumice blocks of WRA were broken into smaller pieces using a clean rock hammer and then lightly crushed with a mortar and pestle. Both the crushed pumice and finer grained tephra samples were sieved to between 0.25 and 0.125 mm and then separated using magnetic methods. The resulting glass and Fe-Ti oxide splits were mounted in epoxy blocks and polished. Major element analyses were performed at University of Toronto using a Cameca SX-50 electron probe microanalyzer (EPMA). Glasses were analyzed at 15 kV accelerating voltage, 5 nA beam current, and a 5 μm beam diameter (Table 4; Table S2 in the Supplemental File [see footnote 1]). Standardization was achieved using mineral and glass standards. Variation between different analytical runs was monitored with the obsidian standard UA5831 and the Old Crow tephra glass shards. Data for Old Crow tephra are reported in Table 4. Fe-Ti oxide minerals were analyzed at 15 kV accelerating voltage, either 15 or 25 nA beam current, and a 1 μm beam diameter (Tables 2 and 3; Tables S3 and S4 in the Supplemental File [see footnote 1]). Standardization was achieved using mineral and synthetic oxide standards. Variation between different analytical runs was monitored with repeated analyses of the Elba hematite, synthetic TiO2, and magnetite and ilmenite from Old Crow tephra (Preece et al., 2011b). FeO and Fe2O3 were calculated using the method of Carmichael (1967). Only Fe-Ti oxide grains with attached glass rinds were analyzed to avoid mineral inclusions and detrital grains. Wherever possible, analyses were performed near the rims of touching magnetite and ilmenite grains. The titanomagnetite analyses reported in this contribution for sample 4 were run at the same time as the ilmenite reported by Richter et al. (1995). Magmatic temperatures and oxygen fugacities were estimated from coexisting titanomagnetite and ilmenite compositions by the method of Ghiorso and Evans (2008), a calibration that includes thermodynamic data for low titanium compositions like those from the WRA (Table S5 in the Supplemental File [see footnote 1]). In all cases, the reported pairs fulfill the permissive equilibrium constraints of Bacon and Hirschmann (1988). In most cases, the Fe-Ti oxide pairs are individual grains in physical contact, except for analyses of the pumice clasts in samples 4, 5, and 6, where this information was not recorded. For these samples the average titanomagnetite and ilmenite compositions were used.

Attempts were made to analyze glass, magnetite, and ilmenite for all samples. It was not possible to analyze the glass in sample 5 because large numbers of microlites are present and all electron-probe analyses were glass-mineral composites. Only a few samples from the Sixtymile River Area (13–22 in Fig. 2 and Table 1) were selected for Fe-Ti oxide analysis.

Individual glass shards were analyzed for trace element contents using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) at Aberystwyth University using a Coherent GeoLas 193nm Excimer laser system coupled to a ThermoFinnigan Element 2 sector field inductively coupled plasma mass spectrometry following the methods in Pearce et al. (2004, 2007, 2011). Calibration was achieved using NIST 612 reference glasses with 29Si as an internal standard, and the calibration was monitored through analysis of the certified ATHO-G glass (Pearce et al., 1997, 2011) (Table 5; Table S6 in the Supplemental File [see footnote 1]). Samples in the proximal area used individual EPMA analyses of SiO2 for internal standardization of glass shards, while average EPMA SiO2 contents were used for distal samples in the Sixtymile area. The location of glass shards major element analyses had not been recorded for the Sixtymile area samples. Purified bulk glass shards from sample 15 were analyzed for trace element contents using instrumental neutron activation analysis at the University of Toronto following the methods of Barnes and Gorton (1984) (Table 5).

Regional Pb isotope analyses for the WVF, including the dacite lava flow at the summit of Mount Churchill (sample 1) and whole pumice sample 4 (pumice clast b) from the eastern lobe of the WRA, are listed in Table 6. In addition, one Nd isotopic analysis was obtained for glass shards from UA459, a sample of the eastern lobe of the WRA located along Holmes Creek (see Lerbekmo et al., 1975, for location) (Table S7 in the Supplemental File [see footnote 1]). Only averaged Nd and Sr isotopic values were presented in Preece and Hart (2004), and for this reason the complete Nd and Sr isotope results for the WVF are listed in Table S7 in the Supplemental File [see footnote 1]. Both Pb and Nd isotope analyses were performed at the Department of Terrestrial Magnetism, Carnegie Institution of Washington, using a VG-354 multicollector mass spectrometer (Table 6; Table S7 in the Supplemental File [see footnote 1]) following the methods of Walker et al. (1989) and Carlson and Irving (1994). Nd isotope ratios are corrected to 146Nd/144Nd = 0.7219 and reported relative to 143Nd/144Nd = 0.511860 for the La Jolla Nd standard. 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb are corrected for mass fractionation by 0.11%, 0.10%, and 0.30%, respectively, relative to the Pb isotope ratios reported by Todt et al. (1996) for the NBS 981 Pb standard. External 2σ uncertainties based on replicate analyses of these standards are 143Nd/144Nd ± 0.000020, 206Pb/204Pb ±0.024, 207Pb/204Pb ±0.026, 208Pb/204Pb ±0.09. Details of the Sr isotope method were provided in Preece and Hart (2004).

1Supplemental File. Tables S1–S7. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00953.S1 or the full-text article on www.gsapubs.org to view the Supplemental File.