Invited review
150,000 years of loess accumulation in central Alaska
Britta J.L. Jensen a, *, 1
, Michael E. Evans b
, Duane G. Froese a
, Vadim A. Kravchinsky b
a
Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, AB, Canada
b
Department of Physics, University of Alberta, Edmonton, AB, Canada
a r t i c l e i n f o
Article history:
Received 1 September 2015
Received in revised form
1 December 2015
Accepted 2 January 2016
Available online 14 January 2016
Keywords:
Loess
Halfway House
Quaternary
Paleosols
Tephra
Magnetic susceptibility
Alaska
Paleomagnetism
Marine isotope stage 5
a b s t r a c t
The Halfway House site in interior Alaska is arguably the most studied loess deposit in northwestern
North America. The site contains a complex paleomagnetic and paleoenvironmental record, but has
lacked the robust chronologic control that would allow its full potential to be exploited. Detailed reexamination
of stratigraphy, paleomagnetics and tephrostratigraphy reveals a relatively complete marine
isotope stage (MIS) 6 to Holocene record constrained by the Old Crow (124 ± 10 ka), VT (106 ± 10 ka),
Sheep Creek-Klondike (ca. 80 ka), Dominion Creek (77 ± 8 ka) and Dawson (ca. 30.2 cal ka BP) tephras.
We show two well-developed paleosols formed during Marine Isotope Stages (MIS) 5e and 5a, while MIS
5c and 5b are either poorly represented or absent. The new tephrostratigraphy presented here is the
most complete one to date for the late Pleistocene and indicates MIS 5 sediments are more common than
previously recognized. A magnetic excursion within the sediments is identiﬁed as the post-Blake
excursion (94.1 ± 7.8 ka), providing independent age control and adding to the increasing body of evidence
that Alaskan loess is a detailed recorder of variations of the Earth's magnetic ﬁeld over time. A
high-resolution magnetic susceptibility proﬁle placed into this new chronostratigraphic framework
supports the hypothesis that wind-intensity is the main variable controlling ﬂuctuations in susceptibility.
Correlation of the susceptibility record to global marine d18
O records is complicated by highly variable
accumulation rates. We ﬁnd the lowest rates of accumulation during peak warm and cold stages, while
abrupt increases are associated with periods of transition between marine isotope (sub)stages. Building
on previous accumulation models for Alaska, surface roughness is likely a leading variable controlling
loess accumulation rates during transitions and peak cold periods, but the negligible accumulation
during MIS 5e and 5a suggests that loess production was exceedingly low, negating the role of surface
roughness. This interplay of variables leads to optimal conditions for loess accumulation during transitions
between isotope stages, and to a somewhat lesser extent, stadials and interstadials.
© 2016 Elsevier Ltd. All rights reserved.
1. Introduction
Large areas of Yukon and Alaska remained unglaciated during
the Quaternary, but the proximity to major ice sheets has provided
the conditions to create and preserve a loess record spanning more
than 3 million years (e.g. Westgate et al., 1990). In this region,
commonly known as eastern Beringia, loess deposits are often
roughly divided into two main types: valley-bottom and upland
loess. Valley-bottom loess is commonly perennially frozen and may
be interbedded with retransported loess and organic material
related to the growth of syngenetic permafrost and periods of
degradation. Sediments are organic-rich, and alluvial and colluvial
material may be locally present. These valley-bottom deposits tend
to preserve the rich ﬂoral and faunal records for which Beringia is
famous (e.g. Froese et al., 2006, 2009; Guthrie, 1968, 1990; Pewe,
1975a; Pewe et al., 1997, 2009; Zazula et al., 2007). Upland loess
is generally comprised of primary air-fall loess that was, or is,
perennially frozen, but has been relatively ice-poor for most of its
existence. Most primary deposits have experienced periods where
permafrost may have completely thawed (e.g. Pewe, 1975a,b; Pewe
et al., 1997), and these ﬂuctuations have caused some reworking
and erosion through thaw slumping and cryoturbation. Regardless,
this ‘upland’ loess is more comparable to ‘classic’ loess deposits,
such as the Chinese loess plateau, where dry, organic-poor loess
contains numerous paleosols (e.g. Liu et al., 1986; Kukla, 1987;
Beget and Hawkins, 1989; Beget, 2001; Muhs et al., 2003, 2008;
Jensen et al., 2008). Modern loess deposition is limited, but does
* Corresponding author.
E-mail address: bjjensen@ualberta.ca (B.J.L. Jensen).
1
Current address: Royal Alberta Museum, Edmonton, AB, Canada.
Contents lists available at ScienceDirect
Quaternary Science Reviews
journal homepage: www.elsevier.com/locate/quascirev
http://dx.doi.org/10.1016/j.quascirev.2016.01.001
0277-3791/© 2016 Elsevier Ltd. All rights reserved.
Quaternary Science Reviews 135 (2016) 1e23
continue in parts of Alaska, with estimated deposition rates at
0.2e2 mm/year in the Fairbanks area (Pewe, 1955, 1975a).
These extensive deposits, originally referred to as the “Yukon
silts,” have been mapped since the late 1800s (e.g. Russell, 1890;
Spurr and Goodrich, 1898; Gilmore, 1908). The origins of the silt
were unclear, with most workers suggesting a number of different
processes, although by the 1950s, Taber's (1943,1953) hypothesis of
in situ production by frost-shattering was in vogue. However,
subsequent detailed work by his contemporary Troy Pewe (1955)
presented a compelling argument that the silts were aeolian deposits.
Pewe (1955) noted that these silts were thickest along or
near the major silt-rich and glacially-fed rivers of Alaska, and tended
to thin away from them, both laterally and with elevation. Most
importantly, he also noted that the mineralogy of the loess differed
from the underlying bedrock, effectively disproving Taber's in situ
hypothesis. This alternative hypothesis was not without controversy
but was largely accepted not long after it was ﬁrst presented.
Pewe went on to formalize the stratigraphic nomenclature for
these silts in the Fairbanks region (Pewe, 1975b). Valley-bottom retransported
loess was thought to be primarily post-last interglacial
(i.e. marine isotope stage (MIS) 5) in age, and was divided into the
Wisconsinan Goldstream Formation (MIS 4e2) and Holocene
Ready Bullion (MIS 1) Formation. Primary loess deposits, which
tend to mantle mid-slope and upland portions of central Alaska,
can also occur stratigraphically below or laterally to the Goldstream
and Ready Bullion Formations, and were named the Engineer Loess
(Holocene) and Gold Hill Loess (>last interglacial). Undifferentiated
upland primary loess deposits were known simply as the Fairbanks
Loess (e.g. Pewe,1955,1975b; Pewe et al.,1966). Since Pewe's initial
work, detailed geochronologic and paleomagnetic studies have
helped differentiate units present in Fairbanks Loess locales (e.g.
Westgate et al., 1990; Oches et al., 1998; Berger, 2003; Lagroix and
Banerjee, 2004a; Pewe et al., 2009). Present usage of Pewe's
nomenclature ﬁnds the terminology generally limited to the Engineer
(Holocene; MIS 1), Goldstream (Wisconsinan; MIS 4e2) and
Gold Hill (>MIS 5) Formations, but application of the formal names
is based on chronology rather than whether the loess is primary or
secondary, or present in valley-bottoms or slopes, since the spatial
distribution and age of deposits have proven to be more complex
(e.g. Beget, 1990; Berger, 2003; Muhs et al., 2003; Lagroix and
Banerjee, 2004a,b). The Eva Creek Formation is the formal name
Pewe et al. (1997) designated to the forest bed/thaw unconformity
that is thought to represent MIS 5e; sediments attributed to other
stages of MIS 5 have not been clearly identiﬁed in Alaska (e.g. Reyes
et al., 2010a). Terminology in this paper will follow recent applications
of Pewe's (1975b) formal stratigraphic nomenclature,
where divisions are based on the age of the loess rather than
location or sedimentology.
The Halfway House site (64.708 N, 148.503 W) is a mid-slope
loess deposit mapped as Fairbanks Loess by Pewe et al. (1966)
(Fig. 1). Located ~47 km west of Fairbanks, this exposure was
created during the construction of the George Parks Highway in the
1960s, and is one of the most extensively studied loess sites in
Alaska and the Yukon (e.g. Westgate et al., 1983, 1985; Beget and
Hawkins, 1989; Beget, 1990; Beget et al., 1990; Oches et al., 1998;
Preece et al., 1999; Vlag et al., 1999; Berger, 2003; Lagroix and
Banerjee, 2002, 2004a,b; Muhs et al., 2003, 2008). Collectively,
this research indicated that Halfway House is mostly comprised of
the Goldstream Formation, and contains a stratigraphic and
paleomagnetic record that has regional paleoenvironmental signiﬁcance.
However, poor chronologic control has hindered the
interpretation of the record, and made it difﬁcult to compare it to
other regional and global paleoenvironmental records. Here we
present a tephrostratigraphic framework for Halfway House that
provides new chronologic control to this section by identifying
several dated tephra beds from other locations in Yukon and Alaska.
High-resolution paleomagnetic data provide additional age constraints
that support the correlations and new age estimates.
Magnetic susceptibility measurements and detailed stratigraphy
allow direct correlation of our record to previous studies, placing
that research into this new chronologic framework, and provide
new insight into Alaskan loess accumulation models.
2. Previous research
2.1. Tephrostratigraphy
Westgate et al. (1983, 1985) ﬁrst described Halfway House and
identiﬁed two paleosols and tephra beds across the section. The
lowest tephra was correlated to the regionally extensive Old Crow
tephra (Fig. 2; OCt; 124 ± 10 ka; Preece et al., 2011a). Preece et al.
(1999) describe the second tephra, as well as two additional beds,
all stratigraphically above OCt, naming them the Halfway House
(HHt), VT and SD tephra beds. At the time, none of these beds had
independent age estimates. Muhs et al. (2003) re-examined
Halfway House and identiﬁed an “unnamed” tephra approximately
1.5 m below OCt and provided several new ages estimates
for the Halfway House tephra that suggested an early Wisconsinan
age.
2.2. Magnetic susceptibility
Unlike Chinese loess, magnetic susceptibility (c) in Alaskan
loess deposits is lower in paleosols and higher in inorganic loess
(e.g. Beget and Hawkins, 1989; Vlag et al., 1999; Liu et al., 2001).
There are several generally accepted assumptions that can guide
interpretations of changes in magnetic susceptibility in Alaskan
loess. First, the dominant ﬂuctuations in susceptibility reﬂect the
size and concentration of magnetic grains present in the loess,
although experimental studies examining susceptibility suggest
concentration may be more important (e.g. Heider et al., 1996). The
amount and size of magnetic grains present in the loess is driven by
changes in wind intensity, with highest wind intensities entraining
more and/or larger magnetic grains, and carrying them further (e.g.
Beget, 1990, 1996, 2001; Vlag et al., 1999). These highest wind intensities
may be a result of katabatic winds from ice sheets (Muhs
and Budahn, 2006), but may also reﬂect differences in synoptic
climatology during the cold stages (e.g. Mock et al., 1998). These
observations indicate that magnetic susceptibility highs occur
during cold stages with extensive glaciation, while the lows are
associated with intervals of decreased glacier coverage and wind
intensity during interglacials or interstadials (e.g. Beget et al.,1990).
Second, loess in Alaska is largely glaciogenic, produced by
comminution and transported via glacially-fed rivers such as the
Tanana, Nenana and Yukon (e.g. Pewe, 1955; Beget, 2001; Muhs
et al., 2003; Muhs and Budahn, 2006). High elevation areas of the
Alaska Range through St. Elias Mountains host extensive modern
glaciers and production and transportation of silt continues today,
albeit in a more limited manner (e.g. Muhs et al., 2003, 2004).
Therefore, accumulation of loess can be continuous through interglacialeglacial
cycles, and paleosols in loess represent periods in
time when loess accumulation was sufﬁciently low that pedogenesis
exceeded loess input (e.g. Muhs and Bettis, 2003; Muhs et al.,
2004).
Further examination of magnetic mineralogy suggests little
pedogenic enhancement of ferrimagnetic content and evidence for
some destruction and alteration of magnetic grains, contributing to
the lower susceptibility within paleosols (Vlag et al., 1999; Liu et al.,
1999, 2001; Beget, 2001). This model of susceptibility variation is
also seen in some Siberian loess deposits (e.g. Chlachula et al.,1998;
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e232
Evans et al., 2003), with some exceptions (Kravchinsky et al., 2008).
Recognition that magnetic susceptibility in Chinese loess
appeared to be highly correlative to benthic foraminiferal d18
O records
(e.g. Heller and Lui, 1986; Kukla et al., 1988) led to a series of
studies on Alaskan loess, including the Halfway House section (e.g.
Beget and Hawkins, 1989; Beget, 1990, 1996; Beget et al., 1990;
Lagroix and Banerjee, 2004a). Assuming a continuous
sedimentation rate, results suggested that the site contained a
complete record of late- Middle Pleistocene to Holocene loess
deposition, and that variations in magnetic susceptibility could be
correlated to marine oxygen isotope records. Observations that
susceptibility proﬁles were similar between same-aged sites
around the Fairbanks region (e.g. Beget, 1990; Vlag et al., 1999), and
statistical tests suggesting signiﬁcant correlation between isotope
Fig. 1. Location of Halfway House (HH). Top: The rectangle marking Fairbanks encompasses the location of Halfway House, and shows the site relative to the landscape during the
last glacial maximum (modiﬁed from Froese et al., 2009). Bottom: Location of HH relative to the village of Ester and geologic units of interest. Map is modiﬁed from Pewe et al.
(1966). Qf- Fairbanks Loess; Qsu-undifferentiated perennially frozen silts; and bc- Birch Creek Schist. Yellow line represents the George Parks Highway.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 3
records and susceptibility (e.g. Beget and Hawkins, 1989), supported
this hypothesis. However, later attempts to reﬁne the
chronology at the site with direct age determinations (e.g. luminescence,
14
C, 10
Be ages) suggested unconformities were present
and much of the record was missing (e.g. Berger, 2003; Muhs et al.,
2003, 2008).
Lagroix and Banerjee (2002, 2004a,b) used anisotropy of
magnetic susceptibility (AMS) to determine regional winddirections
during loess deposition and assess the amount of
post-depositional deformation present at Halfway House; their
continuous proﬁles were developed in trench 1 (Fig. 2). They
found two zones of post-depositional deformation, 30e40 cm
thick, directly above OCt, and associated with the paleosol at
~5e5.5 m (Fig. 3). Implications are that any inclination and
declination values obtained from these intervals will not be reliable.
Using OCt as their age control, they correlated their magnetic
susceptibility curve to SPECMAP. Low susceptibility values directly
above OCt (~3.2 m) were interpreted to represent MIS 5e, while
paleosols at ~5e5.5 m and ~7.5e8 m were correlated to MIS 5a and
a mid-Wisconsinan warm period, respectively. With some modiﬁcation,
this supported Beget's original argument that the record
was relatively complete. Results for wind-directions obtained
from AMS measurements were placed in the SPECMAP correlated
curve, and led them to conclude that wind-directions were predominantly
in the NeS direction during interglacials and NWeSE
during glacials.
2.3. Age control
Oches et al. (1998) were the ﬁrst to attempt to date the site using
thermoluminscence (TL). They made detailed magnetic measurements
and obtained 19 TL dates through trench 1 (Figs. 2 and 3;
Oches et al.,1998; Vlag et al.,1999). The TL ages are stratigraphically
consistent from the base of the Holocene soil to ~10 m, but below
this point the ages become highly variable (Fig. 3). They argued,
despite the age reversal that is present above the paleosol couplet
at ~7.5e8 m, that their ages to ~80 ka, near HHt, are reliable. Berger
(2003) reported a weighted mean TL age of 77.8 ± 4.1 ka for VT,
calculated from one age from Halfway House and two from the
Gold Hill section, placing VT in latest MIS 5a. This age is consistent
with Oches et al. (1998) ages from this depth, but as with them,
Berger also measured a much older sample above the paleosol
couplet.
Muhs et al. (2003, 2008) carried out detailed geochemical analyses
of the loess proﬁle at Halfway House. Using a series of
geochemical proxies developed to identify paleosols and assess the
amount of weathering (e.g. Muhs et al., 1998, 2003, 2008), they
identiﬁed up to seven paleosols, but only two displayed an amount
of development comparable to the modern soil, corresponding to
the paleosol couplet at ~7 m and the paleosol at ~5 m (Fig. 4). Muhs
et al. (2008) concluded, based on non-ﬁnite ages on charcoal from
these paleosols, and the presence of OCt, that the paleosol at
~5.25 m mostly likely formed during the last interglacial, and the
Fig. 2. Halfway House site details. Top: A schematic drawing of Halfway House displaying the most prominent paleosols and continuous tephra beds at the site. Only three trenches
were still clearly visible when the site was revisited in 2007. The dark orange highlights the trenches that were logged and sampled over the course of this study. Bottom: A photo of
the ~south facing slope of Halfway House, the trenches are on the south-east facing cut, which is heavily vegetated with birch, making it difﬁcult to spot the trenches from the
highway. For scale, trench 1 is approximately 12.5 m in height.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e234
Fig. 3. Stratigraphic log of Halfway House with magnetic susceptibility (c) measurements (this study) and ages obtained from previous work. Stratigraphy and c correlate well
between studies, allowing accurate placement of ages. No asterisk ¼ Oches et al. (1998); one asterisk ¼ Muhs et al. (2008); two asterisks ¼ Berger (2003); three asterisks ¼ Auclair
et al. (2007). UA #'s are laboratory numbers assigned to tephra ﬁeld samples. Note the highly variable age estimates below 8 m.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 5
couplet likely formed during an early to mid-Wisconsinan interstadial
(i.e. >50 ka cal yr BP), an assessment consistent with Berger
(2003) and Oches et al. (1998) (Fig. 3).
Auclair et al. (2007) experimented with infrared stimulated
luminescence (IRSL), an optical luminescence method, and tested
their modiﬁcations to the method by dating sediments bracketing
OCt. Their ages of 131 þ 21/À13 ka and 135 þ 21/À13 (Fig. 3) suggested
that their approach was more reliable than TL with loess
older than 80 ka (Auclair et al., 2007). Further testing of the IRSL
method using samples bracketing OCt provided similar ages, but
more importantly suggested that there was no hiatus in loess
deposition over this time (Roberts, 2012). Jensen et al. (2011)
collected samples closely bracketing VT tephra to apply recent
advances in the IRSL approach of Auclair et al. (2007). Evidence at
other sites where VT is present indicated that the latest MIS 5a age
assigned to the tephra was likely too young. Using single-aliquot
regenerative-dose (SAR) IRSL, a new age of 106 ± 10 ka was
determined for VT, which is consistent with paleoenvironmental
and age data from correlative sites (Jensen et al., 2011). This new
age suggests that the early mid-Wisconsinan age assigned to the
paleosol couplet was also too young.
3. Methodology
3.1. Stratigraphy
Initial studies by Westgate et al. (1983, 1985) and Beget and
Hawkins (1989) were completed while the road cut was still well
exposed. Later research required excavation of several trenches as
the section became re-vegetated (Fig. 2). In 2007, we re-excavated a
3 m section in trench 2 to collect bracketing luminescence samples
around VT (c.f. Jensen et al., 2011). This trench was chosen as it had
the best exposure of the tephra and was the location where Berger
(2003) collected his samples. All of trenches 2 and 3 were reexcavated
in 2010 for high-resolution paleomagnetic and tephra
sampling. The upper section of trench 2 does not have a modern
soil proﬁle, thus to ensure a continuous sampling proﬁle to the
present, the upper ~3 m of trench 1 was also re-excavated and
sampled.
3.2. Paleomagnetic sampling and analyses
Magnetic susceptibility and remanence samples were collected
by pushing 2.5 cm diameter plastic cylinders horizontally into the
vertically excavated trenches at 5 cm intervals. Sample orientation
was measured with a Brunton compass, set for local declination. A
total of ~300 samples were collected. Remanence measurements
were made at the University of Alberta by a horizontal 2G Enterprises
DC-squid cryogenic magnetometer equipped with an alternating-ﬁeld
(AF) demagnetizer. Samples were progressively
demagnetized using up to 16 steps and a maximum ﬁeld of 100 mT.
Results were interpreted with the aid of orthogonal vector
component plots and principal component analyses using paleomagnetic
software PMGSC developed by Enkin et al. (2000) Magnetic
susceptibility measurements were made on a Bartington MS-
2 susceptibility meter in three orthogonal orientations (x, y, z),
averaged and normalized by mass; both low (0.465 kHz) and high
frequency (4.65 kHz) measurements were made, with the former
plotted on Figs. 3, 11, 13 and 14 (all measurements available in
Supplementary Data).
3.3. Tephra analyses
Tephra samples were sieved into multiple size fractions; those
with the most abundant glass were utilized for analysis. The majority
of samples collected in this study, with the exception of OCt,
VT and HHt, were small and ﬁne, and the size fraction used was
45e75 mm. Details on preparation methods can be found in Jensen
Fig. 4. Selected geochemical proﬁles modiﬁed from Muhs et al. (2003, 2008). A,B- Depletion of Na2O and SiO2 track the weathering of plagioclase, reﬂected in the plagioclase:quartz
(P:Q) ratio, conﬁrming a relatively signiﬁcant degree of weathering associated with the modern soil, the paleosol couplet at ~7 m, and the paleosol at ~5 m. C- Loess at Halfway
House appears to be sourced mostly from Tanana river silt, but some input from Nenana and Yukon rivers is probable (Muhs and Budahn, 2006). D- Since clay abundance tracks silt
abundance, the variations are likely the results of direct deposition because silt is not a product of pedogenesis. Therefore, peaks of silt and clay within paleosols represent lower
wind intensities resulting in deposition of ﬁner material. EÀ Only the most prominent paleosols at Halfway House show any notable amount of organic material. F- Phosphorous
accumulates in organic matter that is present in the A/O horizon of a soil, but is depleted in the B-horizon by plants. This “high-low” trend of phosphorous has not been documented
in tundra soils, but is present in the modern soil (i.e. boreal forest) and in the two main paleosols.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e236
et al. (2008, 2011). Major-element geochemistry was determined
on single glass shards with a JEOL 8900 superprobe at the University
of Alberta microprobe laboratory using a 15 KeV voltage, 10 mm
beam diameter, and 6 nA current. A Lipari obsidian (ID 3506) and
OCt were run concurrently with all samples to assure proper calibration
and allow accurate comparison between samples run at
different times. All data presented are normalized to 100% and individual
analyses are available in Supplementary Data.
4. Results
4.1. Stratigraphy
The stratigraphic log is a composite from trenches 1, 2, and 3
(Fig. 3). Trench 3 was logged and sampled from the base of the
section to HHt, and this tephra was used to correlate trench 3 to
trench 2. The top of trench 2 corresponds to ~11 m in the composite
stratigraphic log (Fig. 3). The remainder of the section was logged in
the upper 2.5 m of trench 1.
The basal sands that drape the schist bedrock at Halfway House
(e.g. Westgate et al., 1983) were exposed, but the bedrock was not.
The ﬁrst paleomagnetic sample was collected directly above the
contact between the loess and sand, and from this sample to
~0.70 m, the loess contains small sand beds and concretions. Loess
from the sands to ~1.6 m is tan-brown and appears laminated, with
thin beds of organic-rich silts. Small sand beds and laminations
suggest this basal “loess” may be reworked to some extent,
potentially by water. From 1.6 m to 3.4 m the loess is massive,
inorganic and displays heavy mottling deﬁned by irregularly
oxidized loess. This mottling is a common feature of the drier primary
loess deposits in the region and is likely related to past
permafrost thaw. Beyond the oxidation, colors are more typical of
unaltered loess, generally around 2.5 Y 5/2 (e.g. Muhs et al., 2008).
Similar to Muhs et al. (2003, 2008), no morphological evidence of
paleosol formation was noted around OCt in trench 3 (~2.5 m), in
contrast to the reports of Oches et al. (1998) and Lagroix and
Banerjee (2002) (Fig. 5B) in trench 1.
At ~3.4 m the loess becomes more weathered in appearance,
taking on an orange tone that deepens moving towards the paleosol
at 5 m. By 4.4 m, 0.25e1 cm thick bands of mm-scale silt/clay
concretions are cross-cutting through the loess, and the loess itself
has become friable and carries a deep orange hue. Munsell colors
through this section tend to have hues of 7.5 YRe10 YR with high
chromas and values (Fig. 5D). The paleosol (~5e5.4 m) is expressed
as a cryoturbated organic-rich silt bed comprised of cryoturbated
and weathered humic horizons and charcoal. The upper contact
grades into less weathered loess, but occasional humic lenses are
present. The loess becomes ‘speckled’ approaching HHt at ~6.7 m.
The speckles are generally <1 mm diameter, often circular but occasionally
linear and up to a few centimeters in length, weakly
developed, and appear to be composed of Fe (and/or Mn). Some of
these ‘speckles’ are likely rhizoliths, the remains of grass and shrub
roots that oxidized following permafrost thaw. The loess is mottled
and oxidized around HHt and VT, but remains massive and inorganic
to the base of the paleosol couplet (Fig. 5C). The couplet is
~50 cm thick, with a poorly expressed B-horizon at its base. It is
comprised of a lower, 10e20 cm thick, A-horizon intercalated with
organic-rich silt, overlain by a ~10 cm thick less organic-rich Bhorizon,
and capped by another A-horizon/organic-rich silt bed
that is 20e30 cm thick. Charcoal is abundant, the upper contact is
highly undulatory, and the couplet rises and thickens slightly towards
the right end of the trench (Figs. 3 and 5C). Except for two
oxidized intervals, one ~50 cm above the paleosol couplet, the other
at ~10.5 m, which may in fact be a weak B-horizon, the loess is
massive and inorganic for the remainder of trench 2.
Trench 1 was logged and sampled from the modern soil to
~2.5 m below the surface, the base approximately 10e20 cm below
a series of small tephra samples that correlate to those found at the
top of trench 2 (Fig. 5B). Loess is massive and inorganic into the
base of the modern soil, of which the B-horizon becomes visible
~1 m below the surface. The soil is a complex of several interbedded
O, A and B-horizons, rich in charcoal, with a prominent organic bed
visible ~50e60 cm below the surface (Fig. 5A).
4.2. Revised tephrostratigraphy
Muhs et al. (2003, 2008) label an “unidentiﬁed tephra” below
OCt in their stratigraphic logs, and this bed likely corresponds to a
sample collected at ~1.5 m, a continuous, yellow-white bed up to
0.5 cm thick (Fig. 3; UA 1874). This tephra has been correlated to the
Boneyard tephra, ﬁrst identiﬁed at the Palisades site on the Yukon
River, ~250 km west of Fairbanks, but also present at a new Gold
Hill Loess exposure near Ester (Table 1, Fig. 6; Jensen et al., 2013). It
has distinctive major-element glass geochemistry and morphology,
with relatively high Al2O3 weight percent (wt%) and low FeOt wt%
compared to other YukoneAlaska tephra beds at this SiO2 wt %
range. Glass morphology of the bed consists almost entirely of
blocky, thick-walled pumice that have small evenly distributed
vesicles and abundant phenocrysts and microlites.
OCt (UA 1871), ~1 m above Boneyard tephra, is 20e40 cm thick,
has a sharp lower contact, a heavily cryoturbated upper contact,
and includes pods of tephra up to 20 cm above the main deposit
(Figs. 3 and 5). It is a rhyolitic tephra bed with few phenocrysts and
a glass morphology predominantly consisting of platy glass and
thick-walled pumice (Fig. 6; for a detailed review see Preece et al.,
2011a).
Preece et al. (1999) characterize a thin and discontinuous tephra
within “a dark silt that has been deformed by soliﬂuction”
approximately 3 m above Old Crow and 1.5 m below HHt and VT,
which they subsequently named SD tephra. Despite a tentative
identiﬁcation of this tephra in the same stratigraphic location by
Lagroix et al., 2004, we were unsuccessful in recollecting this
tephra. Given the stratigraphy described by Preece et al. (1999) and
Lagroix et al. (2004), we suggest that this tephra is present within
the paleosol described here at 5e5.4 m.
HH1, HH2 and HH3 were identiﬁed from a single sample (UA
1872) collected on a face that was cleared between trenches 2 and 3
to track HHt. Approximately 50 cm below HHt, it was comprised of
several white, diffuse pods, the largest 0.5 cm thick and 7 cm long.
Glass morphology of the sample is predominantly thick-walled
tricuspate and bubble-wall shards, although conspicuous shards
of frothy pumice are present. Major-element geochemical analyses
reveal that three separate populations are present. HH1 is the
predominant population, comprised of tricuspate and bubblewalled
shards, with a distinct geochemical trend that ranges from
low SiO2 wt% dacite to high SiO2 wt% rhyolite. This tephra has been
correlated to the Snag tephra, found in exposures along the White
River, Yukon Territory (Tables 1 and 2; Fig. 7; Turner et al., 2013).
Supporting this correlation is HH2, the second most abundant
population, which is comprised of frothy pumice. HH2 is
geochemically similar to, but can be distinguished from, the Sheep
Creek-Canyon Creek tephra (Fig. 7; SC-CC; e.g. Westgate et al.,
2008). HH2 has been correlated to a previously unreported
tephra bed found in the same stratigraphic sequence as Snag at the
White River exposure, although as a distinct bed (Jensen, unpublished
data). For ease of discussion, we name this tephra informally
as “White River-unknown 5” (WR-UN5), to remain consistent with
nomenclature of Turner et al. (2013). The HH3 population is
recognized from 5 shards in this sample, but is common as detrital
material in virtually all samples collected stratigraphically above,
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 7
but not below, this level. HH3 has been correlated to samples (UA
1624/1909; Fig. 7) collected ~15e20 cm below VT at Gold Hill IV (for
location details see Evans et al., 2011), where it is a mm-scale bed,
semi-continuous, within heavily oxidized and laminated loess. It is
a high SiO2 wt% rhyolite with distinctly low K2O wt% (Table 1;
Fig. 7). Only three beds of similar composition are known: the
Donjek, Kulukak Bay and Stampede tephras. The Donjek and
Kulukak Bay tephras have higher K2O wt%, and the Kulukak Bay
tephra is above VT (Kaufman et al., 2001; Jensen et al., 2011; Turner
et al., 2013). However, glass major-element geochemistry of HH3
and averages of the Stampede tephra presented by Beget and
Keskinen (1991) are notably similar (Tables 1 and 2; Fig. 7). The
Stampede tephra was ﬁrst documented within a block of silt that is
faulted and folded into a push moraine associated with the Lignite
Fig. 5. Important features of Halfway House. A- Paleomagnetic samples collected at 5 cm intervals in the upper ~2.5 m of trench 1. The modern soil complex consists of several
interbedded A and B- horizons. Scale, knife ¼ 20 cm. B- The top of trench 2, two holes are visible and likely remnants from Oches' TL sampling. The wisps of tephra deﬁned by the
dashed line are part of the HH7/HH8 complex and correlate the two trenches. Scale, measuring tape ¼ 80 cm. C- The paleosol couplet above VT and HHt, the 5 cm paleomagnetic
sampling proﬁle was partially collected through the face with HH4. D- OCt, up to 40 cm thick, is within grey-tan loess, while the loess above OCt to the paleosol (base at dashed line)
is weathered, displaying an orange-hue. Scale, shovel ¼ 1.25 m.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e238
glaciation, and a site comprised of interbedded sand and silt on
Richardson Highway by the Tanana River. The Lignite glaciation is
thought to be correlative to the nearby Canyon Creek vertebrate
locality (Weber et al., 1981; Beget and Keskinen, 1991). The tephra
was subsequently correlated the Goose Bay tephra in the Cook Inlet
region (Reger et al., 1996). Age estimates for the Stampede tephra,
based on stratigraphy, 40
Ar/39
Ar and TL dating, range widely from
>140 ka to ~380 ka (Beget and Keskinen, 1991; Reger et al., 1996;
Beget, 2001). However, recent cosmogenic ages (10
Be) reported by
Dortch et al. (2010) suggest that the Lignite Glaciation may have
had an advance in MIS 5d, and the presence of Bison priscus fossils
at Canyon Creek suggest this site must be Rancholabrean in age,
thus likely MIS 6 (e.g. Westgate et al., 2008). These new age
constraints support the correlation, although analyses with reference
material of the Stampede tephra are still required.
HHt (~6.7 m; UA 1453,1870) and VT (~7 m; UA 1425) are discrete
beds up to 10 and 3 cm thick, respectively. Both tephra beds have
diffuse contacts, but similar to OCt and Boneyard tephra, they can
be traced continuously across the exposure, suggesting minimal
post-depositional disturbance (Fig. 5). Both tephra beds are mainly
rhyolitic, but contain some dacitic shards, although the latter are
more abundant in VT (Fig. 6, Table 1). Glass morphology of both
tephra beds is comprised of bubble-walled and tricuspate shards
with some pumice, although HHt glass is much thicker-walled.
Detailed descriptions and geochemistry of HHt and VT are in
Preece et al. (1999) and Jensen et al. (2011, 2013).
HH4 (UA 1869, UA 2107) was collected at the contact between
the paleosol couplet and overlying loess (Figs. 3 and 5C). The
samples consist of diffuse blebs up to 1 cm thick that were semicontinuous
for 70 cm across the trench. HH4 is very ﬁne, most
glass is <75 m, and consists of frothy pumice. Geochemical analyses
show abundant detrital glass shards that have very low Na2O and
high SiO2 concentrations. This suggests they are heavily weathered,
which would make them particularly susceptible to Na-loss during
analyses. The unweathered shards form a coherent population that
has distinct glass geochemistry and morphology that places it in
the Sheep Creek family (c.f. Westgate et al., 2008). Concurrent analyses
with reference material show that HH4 correlates to Sheep
Creek-Klondike (SCt-K, ca. 80 ka; Westgate et al., 2008; 77e81 ka;
Demuro et al., 2013) (Tables 1 and 2; Fig. 7). SCt-K is a widely
distributed tephra bed in the Klondike region south of Dawson City,
Yukon (Fig. 1), and is consistently found directly above a paleosol
(Sanborn et al., 2006; Schweger, 2003; Westgate et al., 2008).
Thirteen samples were collected above SCt-K; a series of
discontinuous blebs and thin beds of tephra, rarely greater than
0.2 cm thick, were collected at intervals to the top of trench 2
(Figs. 3 and 5B). Several small samples were also collected between
2.2 and 2.5 m below the surface in trench 1. Most samples only
contained glass in the 45e75 m fraction. Analyses revealed abundant
detrital material, and in all samples but those collected around
~11 m (i.e. UA 1915 at the surface in trench 2, and most samples in
trench 1), it was difﬁcult to discern any predominant geochemical
population in any one sample. However, taken collectively, we were
able to isolate several distinct geochemical populations within the
samples. HH5 and HH6 comprise two of the largest populations
that were not correlated to older, known tephra beds, such as HH3
(potential Stampede tephra correlative), OCt and HHt. To test if
contamination during laboratory processes may have mixed the
samples, several from different stratigraphic levels were reprocessed
and analyzed, to obtain the same result. In consideration
of prior stratigraphic knowledge and initial results, reference
samples of the Dominion Creek tephra (DCt; Preece et al., 2000,
2011b) and Dome Ash Bed (DAB; e.g. Pewe, 1975b; Preece et al.,
1999) were analyzed with the re-processed samples. The majorTable
1
Major-element glass geochemistry of tephra beds from Halfway House.
Sample SiO2 TiO2 Al2O3 FeOt MnO MgO CaO Na2O K2O Cl H2Odiff n
Boneyard MEAN 76.15 0.04 14.21 0.60 0.14 0.10 1.04 3.94 3.69 0.10 7.40 45
STDEV 0.28 0.04 0.30 0.22 0.03 0.03 0.08 0.47 0.50 0.05 0.81
Old Crow MEAN 75.47 0.31 12.85 1.76 0.08 0.30 1.51 3.72 3.71 0.28 3.34 18
STDEV 0.18 0.04 0.10 0.06 0.03 0.03 0.06 0.16 0.09 0.02 1.29
HH1 MEAN 71.27 0.64 14.20 3.02 0.05 0.68 2.18 4.14 3.71 0.10 4.29 32
STDEV 3.44 0.36 0.78 1.32 0.04 0.45 1.07 0.32 0.65 0.03 1.90
HH2 MEAN 75.13 0.19 14.35 1.24 0.04 0.36 1.67 4.07 2.92 0.03 5.22 12
STDEV 0.80 0.06 0.53 0.12 0.03 0.09 0.38 0.31 0.22 0.02 2.19
HH3 MEAN 76.78 0.25 13.22 1.50 0.10 0.34 1.72 4.47 1.49 0.16 6.52 59
STDEV 0.42 0.04 0.16 0.06 0.04 0.03 0.08 0.28 0.06 0.03 1.79
HHt MEAN 73.04 0.48 14.16 2.49 0.14 0.54 1.99 5.17 1.86 0.14 2.31 84
STDEV 1.47 0.09 0.44 0.56 0.03 0.16 0.50 0.27 0.11 0.02 1.40
VT MEAN 70.51 0.58 14.83 3.17 0.10 0.69 2.47 4.57 2.92 0.16 2.52 79
STDEV 1.36 0.08 0.41 0.49 0.03 0.20 0.42 0.21 0.21 0.04 1.66
HH4 MEAN 73.07 0.20 15.56 1.46 0.05 0.46 2.10 4.43 2.63 0.04 5.77 29
STDEV 0.69 0.05 0.47 0.13 0.03 0.06 0.17 0.16 0.14 0.02 2.57
HH5 MEAN 70.60 0.69 14.56 3.19 0.07 0.68 2.30 4.33 3.48 0.11 2.55 48
STDEV 1.43 0.17 0.33 0.55 0.03 0.20 0.51 0.30 0.30 0.06 1.59
HH6 MEAN 75.88 0.27 13.04 1.34 0.05 0.19 0.88 3.73 4.53 0.11 6.04 73
STDEV 0.53 0.05 0.23 0.13 0.03 0.05 0.15 0.16 0.13 0.03 1.52
HH7 pop.1 MEAN 78.57 0.12 12.41 0.87 0.06 0.18 1.03 3.27 3.28 0.20 6.29 81
STDEV 0.47 0.04 0.37 0.09 0.03 0.05 0.11 0.24 0.32 0.04 1.05
HH7 pop.2 MEAN 76.79 0.14 13.47 1.04 0.06 0.28 1.68 3.83 2.56 0.15 4.59 25
STDEV 0.34 0.04 0.19 0.09 0.03 0.05 0.13 0.16 0.14 0.04 1.90
HH7 all MEAN 78.15 0.13 12.66 0.91 0.06 0.21 1.18 3.40 3.11 0.19 5.89 106
STDEV 0.88 0.04 0.56 0.12 0.03 0.07 0.30 0.33 0.42 0.04 1.48
HH8 MEAN 74.24 0.28 13.54 2.12 0.08 0.23 1.27 4.44 3.58 0.23 4.60 9
STDEV 0.27 0.05 0.18 0.15 0.02 0.05 0.12 0.37 0.14 0.04 2.19
Data normalized on a water-free basis. FeOt ¼ total Fe as FeO; H2Odiff ¼ water by difference; n ¼ number of analyses.
Boneyard ¼ UA 1874; Old Crow ¼ UA 1871; HH1 ¼ UA 1872; HH2 ¼ UA 1872; HH3 ¼ UA 1872, 1857e69, 1873, 2107.
HHt ¼ UA 1870, 1453; VT ¼ UA 1425; HH4 ¼ UA 1869, 2107; HH5/6 ¼ UA 1857e1868, 1873; HH7 ¼ UA 1901e3, 1915, 1451.
HH8 ¼ UA 1901.
Correlations: HH1 ¼ Snag; HH2 ¼ WRUN5; HH3 ¼ Potential Stampede; HH4 ¼ SCt-K; HH5 ¼ DAB; HH6 ¼ DCt; HH8 ¼ Dawson.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 9
element glass geochemistry permits correlation between HH5 and
DAB, and HH6 and DCt (Table 1; Fig. 8). Additionally, UA 1454, a
single sample collected from the exposed face at Halfway House
(Fig. 2), 4 m below the surface, and 4 m above the bedrock contact,
also correlates to HH5 and DAB. This sample contains no other
populations, and indicates that DAB is present at the site as a primary
bed (Fig. 8). The correlation of DCt to HH6 is supported by the
stratigraphy. At all locations where DCt has previously been
collected, it is consistently found 30e100 cm above SCt-K (e.g.
Preece et al., 2000; Westgate et al., 2008). The stratigraphy of DAB is
less certain; although originally placed within the Eva Creek Formation
(e.g. Pewe et al., 1997), Muhs et al. (2008) show it likely
post-dates MIS 5e, but is associated with non-ﬁnite radiocarbon
ages, thus it remains uncertain if DAB is within later MIS 5 or early
MIS 4 loess.
HH7 was collected 5e10 cm from the surface in trench 2 (UA
1915), a pink-grey bleb up to a 1 cm thick and 5 cm long. This
sample correlates to UA 1451, 1902 and 1903 in trench 1, pink wisps
up to 0.5 cm thick and up to 10 cm long, linking the two trenches
(Fig. 9). HH7 is a high SiO2 wt% rhyolite consisting of two main
populations (Table 1; Fig. 9). HH7 has been correlated to three other
sites around the Fairbanks area; Eva Creek, Gold Hill, and remnants
of the Engineer Creek/Dawson Cut exposures near the junction of
the Steese Highway and Gold Mine Trail Road (Fig. 9; for location
details see Pewe et al., 2009). At all sites this tephra bed is a thin but
semi-continuous wisp, within inorganic loess and generally
1e2.5 m beneath the local surface. Glass morphology is also
consistent across all samples, predominantly comprised of chunky
glass shards and pumice that is rich in phenocrysts and microlites,
complicating analyses. Repeated observations of two geochemical
populations that are on trend with each other in samples from
several sites suggests that HH7 is the product of one complex
eruption. HH7 bears a striking resemblance to both the Chatanika
and EB tephras (Table 2; Fig. 9). Chatanika tephra was formally
named by Pewe (1975b), and is described as a 1e10 mm thick white
vitric ash, consistently found at sites in the upper portions of the
Goldstream formation, with a type section designated at the Chatanika
River, 40 km north of Fairbanks. EB tephra has only been
found at Eva Creek (on “Eva Bench”), near the base of what has been
mapped as Engineer Loess (Pewe et al., 2009). Glass morphology of
HH7, Chatanika, EB are similar, particularly notable is the
Fig. 6. Selected bivariate plots of major-element glass geochemistry of tephra beds
previously reported at Halfway House.
Table 2
Major-element glass geochemistry of reference samples.
Sample SiO2 TiO2 Al2O3 FeOt MnO MgO CaO Na2O K2O Cl H2Odiff n
Snag MEAN 68.17 0.99 14.63 4.30 0.07 1.13 3.12 4.38 3.15 0.08 1.55 39
STDEV 2.88 0.25 0.66 1.09 0.02 0.54 0.99 0.23 0.52 0.02 2.17
WRUN5 MEAN 68.17 0.99 14.63 4.30 0.07 1.13 3.12 4.38 3.15 0.08 1.55 39
STDEV 2.88 0.25 0.66 1.09 0.02 0.54 0.99 0.23 0.52 0.02 2.17
Stampede* MEAN 77.23 0.23 13.30 1.34 e 0.34 1.66 4.30 1.46 0.15 3.97 54
STDEV 0.29 0.06 0.15 0.11 e 0.03 0.08 0.16 0.05 0.04 0.86
SCt-K MEAN 72.87 0.21 15.56 1.46 0.04 0.50 2.21 4.58 2.53 0.04 4.09 59
STDEV 0.46 0.04 0.22 0.11 0.03 0.06 0.15 0.23 0.12 0.02 1.98
DAB MEAN 71.43 0.55 14.46 2.67 0.05 0.50 1.86 4.61 3.76 0.10 3.24 20
STDEV 0.53 0.07 0.19 0.19 0.03 0.06 0.16 0.12 0.10 0.02 1.76
DCt MEAN 75.74 0.28 13.05 1.30 0.04 0.19 0.90 3.82 4.59 0.10 5.79 43
STDEV 0.57 0.06 0.26 0.12 0.03 0.04 0.17 0.13 0.12 0.02 1.82
Chatanika** MEAN 76.90 0.13 13.52 1.01 0.08 0.29 1.65 3.78 2.57 0.15 6.01 7
STDEV 0.13 0.13 0.17 0.06 0.02 0.02 0.15 0.13 0.13 0.02 1.75
EB** MEAN 77.62 0.12 13.06 0.83 0.08 0.17 1.26 3.63 3.01 0.22 6.25 12
STDEV 0.86 0.08 0.58 0.12 0.04 0.07 0.22 0.22 0.23 0.07 2.10
Chatanika MEAN 76.67 0.13 13.47 0.99 0.05 0.29 1.54 3.78 2.90 0.19 6.37 23
STDEV 0.72 0.04 0.67 0.15 0.03 0.13 0.31 0.32 0.52 0.08 1.81
EB MEAN 77.60 0.11 12.98 0.89 0.06 0.23 1.17 3.58 3.20 0.19 6.69 16
STDEV 1.13 0.03 0.86 0.16 0.03 0.13 0.33 0.46 0.51 0.03 0.79
Dawson MEAN 74.08 0.27 13.65 2.05 0.07 0.22 1.24 4.52 3.68 0.22 3.67 39
STDEV 0.26 0.04 0.16 0.07 0.03 0.03 0.06 0.21 0.11 0.03 2.25
Data normalized on a water-free basis. FeOt ¼ total Fe as FeO; H2Odiff ¼ water by difference; n ¼ number of analyses.
*Values from Beget and Keskinen (1991) **Values from Pewe et al. (2009). Snag ¼ UA 1589; WRUN5 ¼ UA 1930, 1928.
SCt-K ¼ UT 40; DAB ¼ UT 745; DCt ¼ UT 1905; Chatnika ¼ UT 2178; EB ¼ UA 357; Dawson ¼ UA 1000.
HH1 ¼ Snag; HH2 ¼ WRUN5; HH3 ¼ Stampede?; HH4 ¼ SCt-K; HH5 ¼ DAB; HH6 ¼ DCt; HH8 ¼ Dawson.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2310
abundance of microlites in all three. However, analyses of reference
EB and Chatanika samples with HH7 were inconclusive. Very little
of Pewe's original material remained, thus acquiring enough analyses
for a truly meaningful comparison between tephras with a
relatively broad compositional range was difﬁcult. Fig. 9 shows that
all three tephra are similar to one another, and while EB plots
relatively well with the lower SiO2 population of HH7, Chatanika
captures the entire geochemical range Nonetheless, while Chatanika
is the more likely correlative, particularly in light of its previously
described stratigraphic context, the geochemical data do
not preclude the possibility that each tephra is unique, or that all
are the same bed.
HH8 (UA 1901) was collected in trench 1 at the same level as
HH7. HH8 was a <0.2 mm thick white wisp. This tephra is also
largely comprised of detrital glass, and some shards share afﬁnities
to older tephra beds at the site. However, the largest single population
is a rhyolite that shares many geochemical and glass
morphological characteristics to the Dawson tephra. Originally
described in the Klondike, Yukon, this well characterized bed is part
of a group of geochemically distinct tephra beds that includes OCt,
Fig. 7. Selected bivariate plots of major-element glass geochemistry of HH1, HH2, HH3 and HH4 and their correlative beds. Top row: HH1, HH2 and reference material from Snag,
WR-UN5, and SC-CC tephra beds. Middle row: HH3 plotted with UA 1909 from Gold Hill IV, and values reported for Stampede tephra (with 2 sigma error bars) from Beget and
Keskinen (1991). Bottom row: HH4 plotted with a SC-K reference sample (UT 40).
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 11
Togiak Bay and DL tephras (e.g. Preece et al., 2011a). Dawson is the
only known bed of this group deposited after MIS 5, around
25,300 14
C yr BP (ca. 30.2 cal ka BP; Demuro et al., 2008). Reanalysis
of HH8 with reference material of Dawson tephra, and its
stratigraphic position within ~2.5 m of the surface of the exposure,
support the correlation (Figs. 3 and 8).
4.3. Loess magnetism
Approximately 300 samples at 5-cm resolution scale were
collected for magnetic susceptibility and natural remnant magnetization
(NRM) measurements. A minimum of six AF steps, and an
average of 11, were used for principal component analyses to obtain
characteristic NRM values (ChRM). Many samples carry a secondary
magnetic overprint that is removed in the ﬁrst few demagnetization
steps. It has scattered directions unrelated to the present-day
ﬁeld at Halfway House, and is probably a viscous remanence acquired
during storage in the laboratory. After removal of the overprint,
orthogonal vector end-point plots reveal monotonic decrease
towards the origin demonstrating the presence of a stable
Fig. 8. Selected bivariate plots of major-element glass geochemistry of HH5, HH6, and HH8 and their correlative beds Top row: HH6 and DCt reference material. Middle row: HH5,
DAB reference material, and UA 1454. DAB reference contains fewer dacitic shards than HH5 and UA 1454. This may be the result fewer analyses and/or the processing of reference
material; during glass ﬂotation (e.g. Jensen et al., 2008), it is possible to lose lower SiO2 wt% shards if the density of the separation liquid is too low (e.g. Froese et al., 2003). Bottom
row: HH8, Dawson reference material and UA 1908, a sample from Gold Hill III.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2312
component that we interpret as a record of the ancient geomagnetic
ﬁeld. Some samples exhibit almost ideal behavior (Fig. 10a).
For the entire collection, maximum angular deviation (MAD) values
are generally very small, with 1st quartile, median, and 3rd quartile
values of 0.6, 1.0, and 1.8, respectively. Eight values lie between
5 and 10. A single outlier of 19.4 was rejected. Demagnetization
behavior of representative loess and paleosol samples is illustrated
in Fig.10. The loess (10b) is more magnetic than the paleosol (10c), a
pattern that is also found in susceptibility (2.4 Â 10À6
m3
/kg for the
loess, 0.94 Â 10À6
m3
/kg for the paleosol). While there has been
some evidence of post-depositional destruction of magnetic minerals
(Liu et al., 2001), these variations largely reﬂect the relative
amount of magnetic material present. Lagroix and Banerjee (2002)
show that the magnetic mineralogy is dominated by magnetite
and/or maghemite, with mass fractions of ~0.3% and ~0.03% in loess
and paleosol, respectively. The directional results and susceptibility
data are shown as stratigraphic proﬁles in Fig. 11.
Inclination values recorded in the paleosol at ~7.75 m show a
systematic shallowing, with corresponding shifts in declination.
MAD values throughout this excursional interval have a mean of
Fig. 9. Selected bivariate plots of major-element glass geochemistry of HH7 and samples from other sites. Top row: Correlation of the uppermost sample in trench 2 to trench 1
(HH7). Middle row: HH7 correlates to other samples in the Fairbanks region: UA 1919 (Gold Hill IV), UA 1907 (Gold Hill III), UA 1457/58 (Eva Creek), UA 1683 (Engineering Creek/
Dawson Cut). Bottom row: Reference samples of Chatanika and EB tephra beds plotted with both populations of HH7.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 13
1.4, and never exceed 2.5 (Fig. 11). Once the low-coercivity
overprint is removed, a single underlying magnetic component is
found in all samples. There is no evidence for two magnetic signals
arising from an original detrital remanence on which is superposed
a later component acquired during pedogenesis. There are two
options: either the paleosol retains an original detrital magnetic
signal, or such a component has been completely lost and replaced
by an entirely pedogenic signal. The second option can be rejected
on the basis of the results reported by Lagroix and Banerjee (2002)
who demonstrate that primary ferromagnetic grains not only survive,
but do so in sufﬁcient quantities to faithfully record prevailing
wind directions throughout most of the time represented at
Halfway Housedincluding the interval spanned by the excursional
directions. We conclude that pedogenetic processes have not substantially
altered the paleomagnetic directions we observe.
Age control provided by bracketing tephra beds, VT
(106 ± 10 ka) and SCt-K (~80 ka), indicate this excursion cannot be
attributed to either the Blake (ca. 120 ka; Singer and Hoffman,
2005) or Laschamp (ca. 41 ka; Singer et al., 2009). An extended
period from ~100 to 80 ka of low paleointensity and increased
productivity in 10
Be has been documented in multiple marine cores
(e.g. Frank et al., 1997; Guyodo and Valet, 1999; Carcaillet et al.,
2004; Channell et al., 2009). This interval also contains a
directional excursion, subsequently given the name “post-Blake”
(e.g. Thouveny et al., 2004). However, until recently, only one record
of a directional excursion had been documented during this
interval, from Lac du Bouchet in the French Massif Central
(Thouveny et al., 1990). A series of transitionally magnetized ﬂows
on Iceland, previously ascribed to the Laschamp (e.g. Levi et al.,
1990; Ferk and Leonhardt, 2009), were recently dated to
94 ± 8 ka (Jicha et al., 2011). Jicha et al. (2011) correlate this
excursion to the post-Blake, but resurrect the name of
Skalamælifell, the original name for the Icelandic excursion before
it was erroneously correlated to the Laschamp (Kristja;nsson and
Gudmundsson, 1980). The virtual geomagnetic poles (VGPs) for
the observed low-inclination directions at Halfway House trace out
a trajectory (Fig.12) towards a group of Icelandic VGPs in northwest
Africa reported by Kristjansson (2003) and conﬁrmed by Ferk and
Leonhardt (2009). Kristjansson (2003) suggests that these divergent
VGPs “possibly” belong to the Skalamælifell excursion. Ferk
and Leonhardt (2009) concur, but suggest a much larger VGP
loop. In this study we refer to the event as the post-Blake in light of
the recent review by Singer (2014) on the Quaternary geomagnetic
instability scale (GITS). The age of this excursion agrees well with
the constraints provided by the dated tephra beds that bracket it.
Magnetic susceptibility was measured on the same samples
-0.05
0
0.05
0.1
0.15
0.2
0.25
0 0.05 0.1 0.15 0.2 0.25 0.3
East,Down
North
4.85m
D=342.5, I=69.1, MAD=0.2
N=14 (5-100 mT)
NRM=0.216 A/m
(a)
-0.05
0
0.05
0.1
0.15
0.2
0 0.05 0.1 0.15 0.2 0.25
East,Down
North
10.25m
D=335.1, I=70.4, MAD=0.5
N=11 (15-100 mT)
NRM=0.174 A/m
(b)
0
0.005
0.01
0.015
0.02
0.025
0 0.005 0.01 0.015 0.02 0.025
East,Down
North
7.80m
D=029.0, I=29.5, MAD=1.3
N=12 (12-100 mT)
NRM=0.0297 A/m
(c)
Fig. 10. Orthogonal vector end-point plots. Numerical data given in each plot are stratigraphic height; declination (D), inclination (I) and maximum angular deviation (MAD) (all in
degrees); number of demagnetization steps (N), range given in brackets; natural remanent magnetization (NRM). The three samples illustrate (a) almost ideal demagnetization
behavior, and examples of (b) loess, and (c) paleosol.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2314
used to determine of inclination and declination data, resulting in a
high-resolution proﬁle that is virtually identical to that reported by
Lagroix and Banerjee (2004a), Oches et al. (1998) and Vlag et al.
(1999) from trench 1. This provides the opportunity to place previous
research within the new chronologic framework (Figs. 3 and
13).
5. Discussion
5.1. Interpreting detrital tephra
Loess at Halfway House contains a background of detrital glass
shards that complicate analyses of small samples. Two small
(~1 mm wisps) samples collected near Boneyard tephra were
comprised of glass shards that do not form any interpretable
geochemical populations, and have no afﬁnity to any of the tephra
present at Halfway House. UA 1872, containing HH1 and HH2 as the
most abundant populations, also contained 5 shards of HH3, which
was later identiﬁed in virtually every sample collected above UA
1872. OCt and HHt are also frequently present as detrital shards in
samples collected above their respective stratigraphic levels.
Moving up the stratigraphic column, new populations appear, for
example, DCt and DAB ~50 cm above SCt-K (Fig. 11). In the case of
DCt in particular, its well-documented stratigraphic relationship
with SCt-K from other locations adds conﬁdence to the stratigraphy
at Halfway House, despite the nature of the tephra samples that
Fig. 11. Stratigraphic log of Halfway House including signiﬁcant tephra beds, as well as inclination, declination, MAD, and magnetic susceptibility (c) proﬁles. Highlighted in yellow
is the post-Blake excursion. Lowest c measurements are associated with paleosols and are highlighted. The exception is the small deviation above OCt, which is not associated with
any morphologically distinct paleosol in trench 3 (this study, Muhs et al., 2003, 2008), but has been described by Lagroix et al. (2004a,b) in trench 1 at a similar level.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 15
comprise it (Fig. 3). Therefore, it seems probable that the ﬁrst
appearance of shards in a sample correlating to a known tephra bed
will provide an approximate maximum age for the loess at that
stratigraphic level, and, at a minimum, all loess above this point
must post-date the age of that tephra.
The presence of detrital glass from reworked tephra at the site
suggests that some of the loess is derived from (locally?) reentrained
sediments. This supports Muhs et al.'s (2003) alternative
hypothesis, that the loess has 10
Be inheritance due to previous
subaerial exposure, to explain some of the problematic 10
Be ages at
this site.
5.2. A chronologic framework for Halfway House
Prior to this research, only four tephra beds had been described
at Halfway House: Boneyard, Old Crow, Halfway House and VT
(Westgate et al., 1983, 1985; Preece et al., 1999; Jensen et al., 2013).
Revised age determinations for OCt and VT have been presented in
recent years, and the Boneyard tephra has a regionally consistent
stratigraphic context that provides insight into its time of
deposition.
The Boneyard tephra is within ~1 m of OCt at all sites where it
has been identiﬁed, although it has not been found in direct
stratigraphic context with Sheep Creek-Fairbanks (SCt-F), another
tephra commonly associated with OCt that has a thermoluminescence
age of 190 ± 20 ka (Berger et al., 1996; Preece et al., 1999;
Westgate et al., 2008; Jensen et al., 2013). However, at the Palisades
(reference site for the Boneyard tephra) and Eva Creek, two
sites where SCt-F is in clear stratigraphic context with OCt, SCt-F is
consistently found 3e6 m below OCt (Preece et al., 1999; Muhs
et al., 2001; Jensen et al., 2013). This juxtaposition suggests that
Boneyard tephra is bracketed by SCt-F and OCt, and was deposited
during MIS 6.
The most recent recalculation of multiple ﬁssion-track ages for
OCt places it roughly within MIS 5e (124 ± 10 ka,104e144 ka at two
sigma range; Preece et al., 2011a). However, Reyes et al. (2010b)
described a site at the Palisades where a primary bed of OCt
buried a vegetated surface. The pollen, plant and insect fossils from
this surface show tundra conditions still existed at the time of
deposition. Further careful documentation of multiple sites around
Alaska and the Yukon where the tephra occurs in association with
MIS 5e deposits show that the tephra clearly predates MIS 5e
(Reyes et al., 2010a). Thus, based on stratigraphic relations, and
within error of revised the ﬁssion-track age estimate; we interpret
Fig. 12. Top: A series of VGP paths from the interval containing the post-Blake compared to those from the Skalamælifell lava ﬂows (modiﬁed from Jicha et al., 2011). Bottom:
Halfway House VPGs from samples collected between 6.95 and 8.10 m describe a path towards the North African VGPs of Ferk and Leonhardt (2009).
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2316
the OCt to have been deposited some time in late MIS 6.
Age control on HHt is provided by VT tephra, which lies directly
above HHt at Halfway House and Gold Hill, and by an MIS 5e forest
bed that underlies it at the Palisades (Preece et al., 1999; Berger,
2003; Jensen et al., 2013). The new SAR-IRSL age for VT of
106 ± 10 ka is consistent with evidence at other sites where the
tephra is found (Jensen et al., 2011). In particular, the correlation of
the Aeolis Mountain tephra (Togiak Bay, SW Alaska; Kaufman et al.,
2001) to VT, present in a coastal bluff with independent chronologic
control and a high-resolution pollen proﬁle, suggest VT was
deposited during a stadial in MIS 5 that likely corresponds to MIS
5d (Kaufman et al., 2001; Jensen et al., 2011).
The correlation of the HH1 and HH2 to Snag tephra and WR-UN5
is supported by their close stratigraphic relationship to VT at
Halfway House. Turner et al. (2013) report Snag tephra at a 5 km
long bluff on the White River, southwestern Yukon. This bluff exposes
a sequence of till, non-glacial deposits, and abundant tephra
beds, including OCt, DCt and Dawson. Snag tephra is a newly
described bed presented in this study, which is found stratigraphically
above OCt and associated MIS 5e deposits, and below a
woody peat unit representing MIS 5a that has DCt (77 ± 8 ka)
overlying it. WR-UN5, although not reported in Turner et al. (2013),
is present in close association with Snag at several exposures at this
site (Turner, pers. comm.).
The identiﬁcation of SCt-K, deposited directly on the surface of
the paleosol couplet, and DCt shards ﬁrst appearing in samples
Fig. 13. Correlation of Lagroix and Banerjee (2004a) wind-direction (A) and susceptibility (B) proﬁles to the susceptibility and stratigraphic proﬁle from this study (C), and the LR04
oxygen isotope stack (D) (Lisieski and Raymo, 2005). The SPECMAP proﬁle (D; black curve) is included for comparison (Imbrie et al., 1984). MIS divisions are based on the LR05
record. kmax ¼ maximum susceptibility axes; cmean of 5 samples collected at each 5-cm interval. Ages from dated tephra beds and the excursion are included on (C), TL ages from
Oches et al. (1998) that are considered the most reliable are included on (B). *This age is from the Dawson tephra, but here represents the maximum age as the tephra is likely
detrital.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 17
~50 cm above SCt-K, supports a late MIS 5 age for the couplet. SCt-K
is widely distributed in the Klondike area of central Yukon. Stratigraphically,
SCt-K is consistently found directly above a forest bed,
and generally 1 m below DCt, which has a glass ﬁssion-track age
of 77 ± 8 ka (e.g. Sanborn et al., 2006; Westgate et al., 2008; Preece
et al., 2011b; Zazula et al., 2011). OSL ages reported by Westgate
et al. (2008) of ~80 ka are in agreement with the age of DCt.
Together, these tephras date the forest bed below SCt-K to MIS 5a.
Schweger (2003) reported a detailed, 9-m long, pollen proﬁle
through the Ash Bend site on the Stewart River, central Yukon, that
shows the progression from boreal forest, to forest tundra, to birch
tundra, to herb tundra. SCt-K is present ~1.5 m above the transition
from boreal forest to forest tundra, and is ﬁrmly within the latter.
This indicates that SCt-K was deposited relatively early in the
transition from full interglacial conditions of MIS 5a (boreal forest)
to full glacial conditions of MIS 4 (herb tundra). The replication of
this well-documented stratigraphic relationship at Halfway House
allows correlation to Ash Bend and revises the age of the paleosol
couplet below these tephra beds from the early to midWisconsinan
to late MIS 5.
Whether the paleosol couplet is entirely attributable to MIS 5a
or is a complex that represents MIS 5c, 5b and 5a is not clear. The
post-Blake excursion is most accurately dated to 94.1 ± 7.8 ka in
Iceland, and most strongly expressed in marine paleointensity records
during MIS 5c (~100e95 ka) (Thouveny et al., 2004, 2008;
Jicha et al., 2011). There is little evidence for an unconformity in
this interval, and Lagroix and Banerjee (2004b) did not ﬁnd evidence
of post-depositional deformation (Fig. 13). The excursion
begins ~15 cm below the lower A-horizon, consistent with the age
control provided by VT, but continues through the couplet, with
inclination values not recovering until ~10 cm below SCt-K. Finding
SCt-K on the surface of the paleosol suggests that it is MIS 5a in age,
but it seems improbable, despite the relatively lengthy duration of
the paleointensity low, that the excursion lasted for >10 ka. What
seems more likely is that the upper portion of the paleosol may
have been partially removed during MIS 5a, or there was a hiatus in
loess deposition during peak MIS 5a, resulting in soil formation on
older loess.
The inevitable question that arises from the presence of the
post-Blake event, is the absence of the Blake event. Although records
of the timing and duration of the Blake event are variable,
most studies place it between ca. 125e115 ka, making it largely
coincident with MIS 5e (e.g. Singer and Hoffman, 2005; Lund et al.,
2006; Channell et al., 2012; Singer, 2014). Inclination measurements
at Halfway House reported by Westgate et al. (1983, 1985)
found a shallowing below OCt, an event that has been reported in
the same stratigraphic position in cores from Imuruk Lake on the
Seward Peninsula and Koyukuk Bluff (KY11) in central Alaska (e.g.
Schweger and Matthews, 1985). We were not able to reproduce
these results at Halfway House. Although originally interpreted as
representing the Blake event, this is unlikely because they are
present below OCt (late MIS 6). However, recent identiﬁcation of an
excursion above OCt and directly below an in situ MIS 5e forest bed
at the Palisades, central Alaska, suggests the Blake is recorded in
Alaskan loess of the appropriate age when it has not been disturbed
or removed by MIS 5 thaw (Jensen et al., 2013). At Halfway House,
the paleosol from 5 to 5.5 m is interpreted to represent peak MIS 5e,
and corresponds to one of two intervals where Lagroix and
Banerjee (2004b) found evidence of post-depositional deformation
(Fig. 13). Muhs et al. (2008) also suggested that this paleosol
may have experienced some erosion, partially truncating the top of
the pedogenic geochemical proﬁle. Thus, the absence of the Blake
at Halfway House may be a result of the disturbance and/or absence
of the loess in which the event would have been recorded.
The next dated point in the stratigraphic column is at ~11 m with
the correlation of HH8 to the Dawson tephra. This tephra is found
extensively throughout the Klondike goldﬁelds (e.g. Westgate et al.,
2000; Froese et al., 2002, 2006). Two grass leaves collected from an
in situ surface buried by this tephra at the Goldbottom site in the
Klondike are dated to 25,410 ± 160 and 25,210 ± 260 14
C yr BP
(Froese et al., 2006). Further Bayesian analysis of these ages and ﬁve
additional radiocarbon dates estimate the time of deposition for
Dawson between 30,433 and 30,032 cal yr BP (Demuro et al., 2008).
Dawson tephra is found above organic-rich deposits that are MIS 3
in age, but plant macrofossil data directly associated with the
tephra suggests that the environment of MIS 2 (i.e. steppe-tundra)
is established by the time of deposition (Zazula et al., 2005, 2007;
Froese et al., 2006). Because the population of Dawson tephra in
HH8 is likely detrital, and no samples were collected below this
level in trench 1, it is not certain if HH8 represents the ﬁrst occurrence
of Dawson at Halfway House. Some insight into this problem
is provided by the correlation of UA 1908 from Gold Hill III (for
location details see Evans et al., 2011) to Dawson tephra (Fig. 8). UA
1908 was collected laterally from, and ~75 cm below, UA 1907. UA
1907 correlates to HH7 and at Gold Hill III is a discrete, mm-scale,
continuous bed. Therefore, it appears that Dawson tephra predates
HH7, which was collected at the same stratigraphic level at
Halfway House.
HH7 has been correlated to Eva Creek, Gold Hill III and IV, and
the Engineering Creek/Dawson Cut locales in this study, and at all
sites it is 1e3 m below the surface within massive inorganic loess.
HH7 potentially correlates to the Chatanika and/or EB tephras,
although the possibility that all could be unique units cannot be
discounted. The stratigraphic context of each tephra does not
clarify this problem, as they are all similar to one another. EB tephra
was collected at the Eva Bench cut, adjacent to the more widely
studied Eva Creek exposure, in what was thought to be the base of
the Holocene Engineer Loess, although there are no independent
ages to conﬁrm this assumption (Pewe et al., 2009). Chatanika
tephra was similarly located in inorganic loess considered late
Pleistocene in age. At its reference locality along the Chatanika
River, Pewe (1975b) reports radiocarbon ages (1 sigma error)
bracketing the Chatanika tephra: a date of 14,760 ± 850 14
C yr BP
from a ground squirrel nest ~2 m below the tephra, and a second
date of 14,510 ± 450 14
C yr on coprolites extracted from another
nest ~1.5 m above the tephra.
5.3. Magnetic susceptibility at Halfway House
As reviewed previously, wind intensity is the main variable that
controls changes in magnetic susceptibility in regions such as
Alaska and Siberia: ferromagnetic grains are more abundant and
larger in loess during times of higher wind-intensity (i.e. glacials),
than during times of lower wind-intensity (i.e. interglacials) (e.g.
Beget and Hawkins, 1989; Beget et al., 1990; Vlag et al., 1999).
Consistent with previous studies, we ﬁnd the lowest values in
the magnetic susceptibility at Halfway House are present in the
modern soil complex, the late MIS 5ce5a paleosol complex, and the
MIS 5e paleosol. Additional short-lived susceptibility lows are
present above OCt (~3.25 m) and at ~10.6 m; though both are better
expressed in Lagroix and Banerjee's (2004a) proﬁle of trench 1
(Fig. 13). The low at ~10.6 m is below the Dawson tephra (ca. 30 ka),
and thus may be associated with a very poorly expressed paleosol
of MIS 3 age; paleosols of this age have been documented at several
sites around Fairbanks (e.g. Hamilton et al., 1988; Muhs et al., 2003,
2008).
The persistent, relatively low values below the MIS 5e paleosol
to ~30 cm above OCt raises the possibility that MIS 5e is represented
by that entire interval (Fig. 13). However, although Muhs
et al. (2003, 2008) geochemically identify a truncated soil around
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2318
the level corresponding to the lowest susceptibility values at
~3.25 m, there is no evidence, morphologically or geochemically,
for other individual paleosols through this interval. This interval is
also comprised of the heavily oxidized, friable loess with clay
lamellae described in this study; Muhs et al. (2008) show that
measured ratios of mobile to immobile elements, an indication of
the degree of weathering, is lowest in the paleosol at ~5 m and
remains notably low throughout this section (Fig. 4). This raises the
possibility that this interval may be a B-horizon related to the MIS
5e paleosol at ~5 m, and that weathering could have played a role in
lowering susceptibility in this interval. Weathering, particularly
when related to poorly drained conditions, can lead to some loss of
magnetic grains within Alaskan loess (Lui et al., 1999, 2001; Beget,
2001). An alternative explanation is revealed by the wind-direction
data reported by Lagroix and Banerjee (2004a); this interval is the
only major interval that shows a change in wind direction (Fig. 13).
If this is the case, the steady low in susceptibility could reﬂect a
changing source, particularly if that source is more distal. In support
of this interpretation, this is the only interval where Muhs et al.
(2008) document a change in the geochemistry of the loess at
Halfway House that they interpret as being driven by an inﬂux of
loess derived from the Nenana River, a more distant source than the
Tanana River (Fig. 4).
The slight variation in geochemistry that is tentatively identiﬁed
as a truncated paleosol by Muhs et al. (2003, 2008) appears to
correspond to the susceptibility low at ~3.25 m, which was correlated
to MIS 5e by Lagroix and Banerjee (2004a). This correlation
does not appear likely, but it is a notable feature in the proﬁle, and
its presence in latest MIS 6 suggests it could be related to the
Zeifen-Kattegat climate oscillation (Seidenkrantz et al.,1996). There
is evidence of “two-step” deglaciation at the termination of MIS 6
that is considered analogous to the Younger Dryas, where warming
at the termination of MIS 2 was interrupted by an abrupt return to
glacial conditions before resuming the transition to interglacial
conditions (e.g. Seidenkrantz et al., 1996). The Zeifen-Kattegat
climate oscillation is documented in pollen records across Yukon
and Alaska; at Imuruk Lake, Koyukuk Bluff, Chi'jees Bluff and Birch
Creek, where spruce pollen appears for a short interval directly
above OCt, but is then absent again until MIS 5e (e.g. Seidenkrantz
et al., 1996; McDowell and Edwards, 2001; Reyes et al., 2010b).
The highest susceptibility values are found below OCt (~1e2 m),
below Dawson tephra (~10e10.5 m), and above Dawson tephra
(~12.25 m) (Fig. 13). Age control provided by the tephra beds implies
that these highs correspond to MIS 6, 4 and 2, respectively.
MIS 6 and 4 are better expressed than MIS 2, and this pattern is also
present in the susceptibility record for the upper ~8 m of Gold Hill
IV. The chronology for this latter correlation is provided via the VT
and HH7 tephra beds (Fig. 15). Although MIS 6 is absent at Gold Hill
IV (a large unconformity underlies VT related to thaw during MIS
5e), the record indicates that MIS 4 is more strongly represented
than MIS 2.
The most striking feature of the overall magnetic susceptibility
proﬁle is how it is predominantly comprised of intermediate
values. This suggests that the majority of accumulation at Halfway
House corresponds to times of moderate wind intensity, which
would likely correspond to stadials, interstadials or transitions
between MIS stages, an assumption supported by the chronology.
These results force the modiﬁcation of existing loess accumulation
models in Alaska.
5.4. Loess accumulation in interior Alaska
Silt in Alaska is produced by glacial comminution, without glaciers
there would be little, if any, loess accumulation. However, the
proximity of interior Alaska to glaciers, both presently and in the
past, also complicates attempts to understand loess accumulation
in the region. This complexity is highlighted by Muhs' et al. (2003)
model for loess accumulation that argues that while production is
greatest during full glacial conditions, accumulation rates during
these times are relatively low. They suggest that this reversal reﬂects
the role of surface roughness. For example, while loess production
during full glacial conditions was high, the landscape was
largely covered by herb or steppe-tundra with a low proﬁle and
limited capacity to capture and hold sediment. In contrast, during
interglacials boreal forest is present, with high surface roughness,
leading to higher rates of accumulation despite lower rates of
production (e.g. Muhs et al., 2004). During interstadials, production
is considered high, but accumulation only moderate (Fig. 15).
Our magnetic susceptibility proﬁle, placed in the new chronology,
allows us to test this model. Peak values of susceptibility,
linked to wind intensity and full glacial conditions are not the
dominant feature of the proﬁle. However, while there is very little
accumulation during MIS 2, consistent with the full glacial scenario
of Muhs’ et al. (2003, 2004) model and their observations from
several equivalent-aged loess localities around Fairbanks, we do
ﬁnd that there is accumulation during MIS 4 and 6. The minimal
amount of loess from MIS 2 is consistent with glacial records in
Alaska that indicate MIS 4 and MIS 6 ice limits, from both the Alaska
and Brooks Ranges, were more extensive than during MIS 2 (e.g.
Briner and Kaufman, 2008; Dortch et al., 2010; Matmon et al.,
2010). This suggests while surface roughness may be the
Fig. 14. Correlation of the Halfway House and Gold Hill IV susceptibility proﬁles. The
correlation is supported by the presence of VT (106 ± 10 ka) and HH7 at both sites.
Dawson tephra (ca. 30 cal yr BP) is not present at Gold Hill IV, but is ~50 cm below HH7
at Gold Hill III, approximately 150 m southwest from Gold Hill IV. Neither DCt
(77 ± 8 ka) or SCt-K (ca. 80 ka) have been found at Gold Hill, but DAB is present at the
Gold Hill Steps section, ~200 m east on the George Parks Highway. This suggests that
the prominent humiﬁed peat unit that is ubiquitous several meters below the surface
across much of Gold Hill is most likely associated with MIS 3, consistent with radiocarbon
dates, but the paleosol below DAB at Gold Hill Steps is likely associated with
MIS 5a rather than an early to mid-Wisconsinan interstadial (e.g. Muhs et al., 2003).
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 19
controlling variable during full glacials, the production of silt is still
important.
We also do not see signiﬁcant loess accumulation during periods
of peak warmth in MIS 5, and, in fact, a hiatus may even be present.
It follows that during peak interglacial warmth, although surface
roughness is high, glaciers are at their minimum extent and silt
production is low, and thus becomes more critical than surface
roughness as the dominant variable controlling accumulation rates.
As noted previously, most of the loess has intermediate susceptibility
values, reﬂecting moderate wind conditions (and
potentially distance from source). These conditions and the new
chronology suggest that the highest rates of accumulation at
Halfway House and the upper-most portion of Gold Hill occur
during transitions between isotope stages, and to a lesser extent,
stadials and interstadials. With glaciers advancing or retreating at
these times, production rates would have been relatively high, as
well as surface roughness, since transitional environments would
have consisted more of shrub tundra, potentially interspersed with
stands of trees such as birch and aspen, and even sparse spruce (e.g.
Anderson and Lozhkin, 2001; Brubaker et al., 2001, 2005;
Schweger, 2003). This represents optimal conditions for loess
accumulation, where loess production and surface roughness are
both relatively high.
There are factors that should be considered when applying this
revised model:
(1) It is most applicable to the thicker loess deposits that are
found more “distally” from their glacial sources. For example,
loess deposits closer to source (e.g. in the Matanuska, Tanana,
Delta or Nenana river valleys 10 s of kms from or in the
Alaska Range, near present or past glacial limits) tend to
preserve shorter records and may, in fact, only preserve
interglacial loess. This may be the result of ice covering the
area during an advance, or the high wind intensity and low
height of the vegetative surface during glacials resulting in
deﬂation (e.g. Thorson and Bender, 1985; Beget, 2001; Muhs
et al., 2004).
(2) Many loess deposits ﬁll topographic lows (i.e. valleys) or are
on the lee-side of bedrock highs. For example, the
conspicuously thick Gold Hill loess is deposited over a
particularly low bedrock hill (e.g. Pewe et al., 2009). However,
once the deposit reaches the elevation of the local
surface (ﬁlling all available accommodation space), further
deposition is unlikely and deﬂation may even occur. For
example, the upper 8 m of late Pleistocene to Holocene loess
at the top of Gold Hill IV is likely partially due to the lowering
of the local surface by the MIS 5e thaw unconformity (Eva
Creek Formation). This 8 m represents only ~100,000 years in
a deposit that spans about 3 million years (e.g. Westgate
et al., 1990). Therefore the upper 8 m of the section only
covers 3% of the time represented, but makes up 13% of the
section. This illustrates how changes in accommodation
space can also enhance or minimize the deposition of loess in
this region.
During MIS 5 there are likely short hiatuses in deposition and/or
partial removal of paleosols (e.g. the decapitation of the MIS 5e
paleosol) at Halfway House. Muhs et al. (2003) report 10
Be accumulation
data that indicated the presence of an unconformity. They
suggested that it was most likely located near OCt, which is not
likely given the revised chronology, but doesn't preclude an unconformity
higher in the section. However, although there appears
to be a slow-down in accumulation in the upper ~1.5 m of the
section, and the MIS 3 paleosol is not as well expressed as it is at
Gold Hill (possibly partially truncated); there is little evidence of a
major unconformity. Overall, the chronology and magnetic susceptibility
data presented here shows that Halfway House is
somewhat unusual in comparison to many other loess exposures in
this region in that it appears to be relatively continuous (e.g. Jensen
et al., 2008; Pewe et al., 2009). While there is a no evidence for any
prominent unconformities, the lateral continuity of the site cannot
be fully examined, and it should be noted that the Holocene soil is
not present in trench 2, although that may be the result of disturbance
during the construction of the original road cut. Regardless,
the relative continuity of the section and evidence for loess accumulation
in the Holocene suggests that rates of loess deposition
and changes in magnetic susceptibility at this site are more
reﬂective of broad environmental and climatic changes rather than
Fig. 15. Comparison of Muhs et al. (2003) idealized loess accumulation model, roughly based on Halfway House stratigraphy, to the one developed in this study using magnetic
susceptibility and new age data as guides. The models are similar but peak susceptibility measurements indicate full glacial loess is present, there may be a hiatus during peak
interglacial warmth, and that more loess appears to accumulate during transitions between isotope stages than at any other time.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2320
variations in local conditions. If so, this site represents an ideal
location to guide interpretations of other loess deposits in the
region.
5.5. Marine isotope stage 5 in Alaska
In the Yukon there are multiple sites that preserve MIS 5a and 5e
deposits, although only two sites are known with both in sequence
(e.g. Westgate et al., 2008; Reyes et al., 2010a; Turner et al., 2013).
At these sites, if other MIS 5 sediments are present, it is usually a
loess unit containing one or more of series of tephra beds, such as
Snag, Woodchopper Creek or Donjek, that identify it as MIS 5d (e.g.
Turner, 2013). However, no evidence has been found of sediments
that might have been deposited during MIS 5c or 5b.
Last interglacial deposits beyond MIS 5e are not clearly recognized
in Alaska, but this could be the result of unreliable chronologies.
For example, the Gold Hill Steps section, which has been
examined along with Halfway House (e.g. Vlag et al., 1999; Muhs
et al., 2003, 2008; Lagroix and Banerjee, 2004a,b), contains a
similar susceptibility record and comparable stratigraphy in its
upper ~8 m. Below a paleosol complex dated by Muhs et al. (2003,
2008) to the mid-Wisconsin (MIS 3), there is a peak in susceptibility
that appears to correlate to the MIS 4 peak at Halfway House. Below
this peak there is ~2.5 m of loess with intermediate susceptibility
values to a well-developed paleosol, which is overlain by DAB. The
correlation of Gold Hill Steps to Halfway House is supported by the
correlation of Halfway House to the upper 8 m of the Gold Hill IV
exposure (GHIV; Fig. 14), which is only ~200 m from the steps. The
stratigraphic and magnetic susceptibility records between the upper
8 m of GHIV and Gold Hill Steps are virtually indistinguishable,
except for the absence of DAB at GHIV (Vlag et al., 1999, Fig. 14).
Muhs et al. (2008) suggest the paleosol below DAB at Gold Hill
Steps is representative of the last interglacial, however, in light of
the evidence at Halfway House and Gold Hill IV, it seems more
likely that it was formed during MIS 5a. Probable MIS 4 deposits
and DAB were also identiﬁed at Eva Creek (Muhs et al., 2001).
Therefore, MIS 5 sediments, in addition to MIS 5e, may be more
common in Alaska than has previously considered.
6. Conclusions
The Halfway House site is one of the most studied loess sections
in Alaska, but interpretations of those studies have been limited by
poor chronological control. Systematic re-examination of the
tephrostratigraphy has identiﬁed several dated tephra beds not
previously recognized at the site: SCt-K (ca. 80 ka), DCt (77 ± 8 ka)
and Dawson (~30 cal yr BP). The presence of the post-Blake
excursion (94.1 ± 7.8 ka) provides an independent age at the site,
and supports the chronology provided by SCt-K and VT
(106 ± 10 ka), which bracket the excursion. These ages, and the
presence of Old Crow (124 ± 10 ka) and Boneyard tephras, indicate
that Halfway House has a relatively complete MIS 6 to Holocene
record, with little evidence of major unconformities, although there
may be minor truncations and hiatuses, particularly through MIS 5.
This revised chronology also indicates the presence of two major
paleosols that formed during MIS 5 (Fig. 13). The identiﬁcation of
Snag, WR-UN5, SCt-K, and DCt is the ﬁrst documentation of these
tephra beds outside of the Yukon, and indicates that they are
important regional markers for MIS 5d (Snag, WR-UN5) and MIS 5a
(SCt-K, DCt). The appearance of DAB glass shards above the late MIS
5 paleosol at Halfway House, and the implication that the paleosol
below DAB at the Gold Hill Steps site correlates to this paleosol,
suggests that this tephra bed is of similar age to DCt (ca. 77 ka), and
that MIS 5a may be more represented in Alaska than previously
considered.
These new ages, when placed on the magnetic susceptibility
proﬁle, show that loess accumulation at Halfway House is highest
during transitions between MIS stages, interstadials and stadials,
with low loess accumulation during peak cold and peak warm intervals.
This is partially consistent with the loess accumulation
model presented by Muhs et al. (2003), where loess production and
accumulation are not positively correlated, and surface roughness
is the leading variable controlling accumulation rates. In contrast to
their model, we ﬁnd that during full glacial conditions there is
evidence for some accumulation (albeit generally at lower rates),
but also ﬁnd a lack of evidence for high accumulation rates during
peak interglacial times. We modify the model to suggest that loess
production is the most important variable controlling accumulation
during times of extensive glacier retreat (e.g. MIS 5e), and that the
greatest amount of accumulation is during times when loess production,
surface roughness, and wind intensity are all relatively
moderate.
The complexity of highly variable accumulation rates dependent
on the interplay of wind intensity, loess production, surface
roughness and location of the site, demands the development of
robust chronologies when attempting to link Alaskan magnetic
susceptibility proﬁles to global d18
O curves or other paleoclimate
and paleomagnetic records. However, once the chronology is
established, the almost ideal paleomagnetic characteristics of these
loess deposits shows that they are exceptional archives that provide
important insights into past regional and global paleomagnetic and
paleoenvironmental variations.
Acknowledgments
This research was funded by the Natural Sciences and Engineering
Research Council of Canada (D.G.F.), and grants from the
Canadian Circumpolar Institute and Northern Scientiﬁc Training
Program (B.J.L.J.). B.J.L.J. acknowledges scholarship support from
NSERC, Alberta Ingenuity, and the Killam Trusts. Excellent assistance
with the hard labour and patience required for the ﬁeldwork
were provided by H. Dunning and J. Pumple. We thank F. Lagroix for
generously providing access to ﬁgures and photos on her research
at Halfway House that helped us properly pinpoint the sections
referred to in previous research, and S.J. Preece and J.A. Westgate
for providing reference material for EB, Chatanika, Dominion Creek
and DAB tephra beds. B.J.L.J. also thanks Dr. Anders Carlson for
providing the space and time to complete her PhD while a visiting
scientist at the University of Wisconsin-Madison, where the ﬁrst
iteration of this study appears. We also thank the reviewers, France
Lagroix and Dan Muhs for their constructive comments that
resulted in an improved manuscript.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.quascirev.2016.01.001.
References
Anderson, P.M., Lozhkin, A.V., 2001. The stage 3 interstadial complex (Karginskii/
middle Wisconsinan interval) of Beringia: variations in paleoenvironments and
implications for paleoclimatic interpretations. Quat. Sci. Rev. 20, 93e125.
Auclair, M., Lamothe, M., Lagroix, F., Banerjee, S., 2007. Luminescence investigation
of loess and tephra from Halfway House section, Central Alaska. Quat. Geochronol.
2, 34e38.
Beget, J., 1990. Middle Wisconsinan climate ﬂuctuations recorded in central Alaskan
loess. Geogr. Physique Quaternaire 44, 3e13.
Beget, J., 1996. Tephrochronology and paleoclimatology of the last interglacialglacial
cycle recorded in Alaskan loess deposits. Quat. Int. 34, 121e126.
Beget, J.E., 2001. Continuous Late Quaternary proxy climate records from loess in
Beringia. Quat. Sci. Rev. 20, 499e507.
Beget, J.E., Hawkins, D.B., 1989. Inﬂuence of orbital parameters on Pleistocene loess
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 21
deposition in central Alaska. Nature 337, 151e153.
Beget, J.E., Keskinen, M., 1991. The Stampede tephra: a middle Pleistocene marker
bed in glacial and eolian deposits of central Alaska. Can. J. Earth Sci. 287,
991e1002.
Beget, J.E., Stone, D.B., Hawkins, D.B., 1990. Paleoclimatic forcing of magnetic susceptibility
variations in Alaskan loess during the late Quaternary. Geology 18,
40e43.
Berger, G.W., Pewe, T.L., Westgate, J.A., Preece, S.J., 1996. Age of Sheep Creek tephra
(Pleistocene) in central Alaska from thermoluminescence dating of bracketing
loess. Quat. Res. 45, 263e270. http://dx.doi.org/10.1006/qres.1996.0027.
Berger, G.W., 2003. Luminescence chronology of late Pleistocene loess-paleosol and
tephra sequences near Fairbanks, Alaska. Quat. Res. 60, 70e83. http://
dx.doi.org/10.1016/S0033-5894(03)00060-7.
Briner, J.P., Kaufman, D.S., 2008. Late Pleistocene mountain glaciation in Alaska: key
chronologies. J. Quat. Sci. 23, 659e670. http://dx.doi.org/10.1002/jqs.1196.
Brubaker, L.B., Anderson, P.M., Hu, F.S., 2001. Vegetation ecotone dynamics in
Southwest Alaska during the Late Quaternary. Quat. Sci. Rev. 20, 175e188.
Brubaker, L.B., Anderson, P.M., Edwards, M.E., Lozhkin, A., 2005. Beringia as a glacial
refugium for boreal trees and shrubs: new perspectives from mapped pollen
data. J. Biogeogr. 32, 833e848. http://dx.doi.org/10.1111/j.1365-
2699.2004.01203.x.
Carcaillet, J., Bourles, D.L., Thouveny, N., Arnold, M., 2004. A high resolution
authigenic 10
Be/9
Be record of geomagnetic moment variations over the last 300
ka from sedimentary cores of the Portuguese margin. Earth Planet. Sci. Lett. 219,
397e412. http://dx.doi.org/10.1016/S0012-821X(03)00702-7.
Channell, J.E.T., Xuan, C., Hodell, D.A., 2009. Stacking paleointensity and oxygen
isotope data for the last 1.5 Myr (PISO-1500). Earth Planet. Sci. Lett. 283, 14e23.
http://dx.doi.org/10.1016/j.epsl.2009.03.012.
Channell, J.E.T., Hodell, D.A., Curtis, J.H., 2012. ODP Site 1063 (Bermuda Rise)
revisited: oxygen isotopes, excursions and paleointensity in the Brunhes Chron.
Geochem. Geophys. Geosyst. 13 http://dx.doi.org/10.1029/2011GC003897.
Q02001.
Chlachula, J., Evans, M.E., Rutter, N.W., 1998. A magnetic investigation of a late
Quaternary loess/palaeosol record in Siberia. Geophys. J. Int. 132, 128e132.
Demuro, M., Roberts, R.G., Froese, D.G., Arnold, L.J., Brock, F., Ramsey, C.B., 2008.
Optically stimulated luminescence dating of single and multiple grains of
quartz from perennially frozen loess in western Yukon Territory, Canada:
comparison with radiocarbon chronologies for the late Pleistocene Dawson
tephra. Quat. Geochronol. 3, 346e364. http://dx.doi.org/10.1016/
j.quageo.2007.12.003.
Demuro, M., Arnold, L.J., Froese, D.G., Roberts, R.G., 2013. OSL dating of loess deposits
bracketing Sheep Creek tephra beds, northwest Canada: Dim and problematic
single-grain OSL characteristics and their effect on multi-grain age
estimates. Quat. Gechronol. 15, 67e87. http://dx.doi.org/10.1016/
j.quageo.2012.11.003.
Dortch, J.M., Owen, L.A., Caffee, M.W., Li, D., Lowell, T.V., 2010. Beryllium-10 surface
exposure dating of glacial successions in the Central Alaska Range. J. Quat. Sci.
25, 1259e1269. http://dx.doi.org/10.1002/jqs.1406.
Enkin, R., Wuolle, K., McCann, C., Carretero, M., Voroney, M., Baylis, T., Morton, K.,
Jaycock, D., Baker, J., Beran, L., 2000. PMGSC, Paleomagnetism Data Analysis.
Version 4.2. www.pgc.nrcan.gc.ca/tectonic/enkin.htm.
Evans, M.E., Rutter, N.W., Catto, N., Chlachula, J., Nyvlt, D., 2003. Magnetoclimatology:
teleconnection between the Siberian loess record and North
Atlantic Heinrich events. Geology 31, 537e540.
Evans, M.E., Jensen, B.J.L., Kravchinsky, V.A., Froese, D.G., 2011. The Kamikatsura
event in the gold Hill loess, Alaska. Geophys. Res. Lett. 38 http://dx.doi.org/
10.1029/2011GL047793. L13302.
Ferk, A., Leonhardt, R., 2009. The Laschamp geomagnetic ﬁeld excursion recorded in
Icelandic lavas. Phys. Earth Planet. Inter. 177, 19e30. http://dx.doi.org/10.1016/
j.pepi.2009.07.011.
Frank, M., Schwarz, B., Baumann, S., Kubik, P.W., Suter, M., Mangini, A., 1997. A 200
kyr record of cosmogenic radionuclide production rate and geomagnetic ﬁeld
intensity from 10
Be in globally stacked deep-sea sediments. Earth Planet. Sci.
Lett. 149, 121e129.
Froese, D., Westgate, J., Preece, S., Storer, J., 2002. Age and signiﬁcance of the late
Pleistocene Dawson tephra in eastern Beringia. Quat. Sci. Rev. 21, 2137e2142.
Froese, D.G., Smith, D.G., Westgate, J.A., Ager, T.A., Preece, S.J., Sandhu, A., Enkin, R.J.,
Weber, F., 2003. Recurring middle Pleistocene outburst ﬂoods in east-central
Alaska. Quat. Res. 60, 50e62. http://dx.doi.org/10.1016/S0033-5894(03)00090-
5.
Froese, D.G., Zazula, G.D., Reyes, A.V., 2006. Seasonality of the late Pleistocene
Dawson tephra and exceptional preservation of a buried riparian surface in
central Yukon Territory, Canada. Quat. Sci. Rev. 25, 1542e1551. http://dx.doi.org/
10.1016/j.quascirev.2006.01.028.
Froese, D.G., Zazula, G.D., Westgate, J.A., Preece, S.J., Sanborn, P.T., Reyes, A.V.,
Pearce, N.J.G., 2009. The Klondike goldﬁelds and Pleistocene environments of
Beringia. GSA Today 19, 4. http://dx.doi.org/10.1130/GSATG54A.1.
Gilmore, C.W., 1908. Smithsonian exploration in Alaska in 1907 in search of Pleistocene
fossil vertebrates. Smithson. Misc. Collect. 51, 38.
Guthrie, R.D., 1968. Paleoecology of the large-mammal community in interior
Alaska. Am. Midl. Nat. 79, 346e363. http://dx.doi.org/10.2307/2423182.
Guthrie, R.D., 1990. Frozen Fauna of the Mammoth Steppe: the Story of Blue Babe.
University of Chicago Press, Chicago, p. 323.
Guyodo, Y., Valet, J.P., 1999. Global changes in intensity of the earth's magnetic ﬁeld
during the past 800 kyr. Nature 399, 249e252.
Hamilton, T.D., Craig, J.L., Sellman, P.V., 1988. The Fox permafrost tunnel: a late
Quaternary geologic record in central Alaska. Geol. Soc. Am. Bull. 100, 948e969.
Heider, F., Zitzelsberger, A., Fabian, K., 1996. Magnetic susceptibility and remanent
coercive force in grown magnetite crystals from 0.1 mm to 6 mm. Phys. Earth
Planet. Inter. 93, 239e256.
Heller, F., Lui, T., 1986. Palaeoclimatic and sedimentary history from magnetic
susceptibility of loess in China. Geophys. Res. Lett. 13, 1169e1172.
Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G.,
Prell, W.L., Shackleton, N.J., 1984. The orbital theory of Pleistocene climate:
support from a revised chronology of the marine d18
O record. In: Berger, A.L.
(Ed.), Milankovitch and Climate, vol. 1D. Reidel Publishing Company, Norwell,
MA, pp. 269e305.
Jensen, B., Froese, D., Preece, S., Westgate, J., Stachel, T., 2008. An extensive middle
to late Pleistocene tephrochronologic record from east-central Alaska. Quat. Sci.
Rev. 27, 411e427. http://dx.doi.org/10.1016/j.quascirev.2007.10.010.
Jensen, B.J.L., Preece, S.J., Lamothe, M., Pearce, N.J.G., Froese, D.G., Westgate, J.A.,
Schaefer, J., Beget, J., 2011. The Variegated (VT) tephra: a new regional marker
for middle to late marine isotope stage 5 across Yukon and Alaska. Quat. Int.
246, 312e323. http://dx.doi.org/10.1016/j.quaint.2011.06.028.
Jensen, B.J.L., Reyes, A.V., Froese, D.G., Stone, D.B., 2013. The Palisades is a key
reference site for the middle Pleistocene of eastern Beringia: new evidence
from paleomagnetics and regional tephrostratigraphy. Quat. Sci. Rev. 63,
91e108. http://dx.doi.org/10.1016/j.quascirev.2012.11.035.
Jicha, B.R., Kristjansson, L., Brown, M.C., Singer, B.S., Beard, B.L., Johnson, C.M., 2011.
New age for the Skamælifell excursion and identiﬁcation of a global geomagnetic
even in the late Brunhes chron. Earth Planet. Sci. Lett. 310, 509e517.
http://dx.doi.org/10.1016/j.epsl.2011.08.007.
Kaufman, D., Manley, W., Wolfe, A., Hu, F., Preece, S., Westgate, J., Forman, S., 2001.
The last interglacial to glacial transition, Togiak Bay, southwestern Alaska. Quat.
Res. 55, 190e202.
Kravchinsky, V.A., Zykina, V.S., Zykin, V.S., 2008. Magnetic indicator of global
paleoclimate cycles in Siberian loess. Earth Planet. Sci. Lett. 265, 498e514.
http://dx.doi.org/10.1016/j.epsl.2007.10.031.
Kristjansson, L., Gudmundsson, A., 1980. Geomagnetic excursion in late-glacial
basalt outcrops in south-western Iceland. Geophys. Res. Lett. 7, 337e340.
Kristjansson, L., 2003. Paleomagnetic observations on Late Quaternary basalts
around Reykjavík and on the Reykjanes peninsula, SW-Iceland. J€okull 52,
21e32.
Kukla, G., 1987. Loess stratigraphy in central China. Quat. Sci. Rev. 6, 191e219.
Kukla, G., Heller, F., Ming, L.X., Chun, X.T., Sheng, L.T., Sheng, A.Z., 1988. Pleistocene
climates in China dated by magnetic susceptibility. Geology 16, 811e814.
Lagroix, F., Banerjee, S.K., 2002. Paleowind directions from the magnetic fabric of
loess proﬁles in central Alaska. Earth Planet. Sci. Lett. 195, 99e112. http://
dx.doi.org/10.1016/S0012-821X(01)00564-7.
Lagroix, F., Banerjee, S.K., 2004a. The regional and temporal signiﬁcance of primary
aeolian magnetic fabrics preserved in Alaskan loess. Earth Planet. Sci. Lett. 225,
379e395. http://dx.doi.org/10.1016/j.epsl.2004.07.003.
Lagroix, F., Banerjee, S.K., 2004b. Cryptic post-depositional reworking in aeolian
sediments revealed by the anisotropy of magnetic susceptibility. Earth Planet.
Sci. Lett. 224, 453e459. http://dx.doi.org/10.1016/j.epsl.2004.05.029.
Lagroix, F., Banerjee, S.K., Jackson, M.J., 2004. Magnetic properties of the Old Crow
tephra: identiﬁcation of a complex iron titanium oxide mineralogy. J. Geophys.
Res. 109 http://dx.doi.org/10.1029/2003JB002678. B01104.
Levi, S., Audunsson, H., Duncan, R.A., Kristjansson, L., Gillot, P.Y., Jakobsson, S.P.,
1990. Late Pleistocene geomagnetic excursion in Icelandic lavas: conﬁrmation
of the Laschamp excursion. Earth Planet. Sci. Lett. 96, 443e457.
Lisiecki, L.E., Raymo, M.E., 2005. A PlioceneePleistocene stack of 57 globally
distributed benthic d18
O records. Paleoceanography 20, 522e533.
Liu, T., Zhang, S., Han, J., 1986. Stratigraphy and paleoenvironmental changes in the
loess of central China. Quat. Sci. Rev. 5, 489e495.
Liu, X., Hesse, P., Rolph, T., Beget, J., 1999. Properties of magnetic mineralogy of
Alaskan loess: evidence for pedogenesis. Quat. Int. 62, 93e102. http://
dx.doi.org/10.1016/S1040-6182(99)00027-0.
Liu, X.M., Hesse, P., Beget, J., Rolph, T., 2001. Pedogenic destruction of ferrimagnetics
in Alaskan loess deposits. Aust. J. Soil Res. 39, 99. http://dx.doi.org/10.1071/
SR99081.
Lund, S., Stoner, J., Channell, J., Acton, G., 2006. A summary of Brunhes paleomagnetic
ﬁeld variability recorded in Ocean Drilling Program cores. Phys. Earth
Planet. Inter. 156, 194e204.
Matmon, A., Briner, J.P., Carver, G., Bierman, P., Finkel, R.C., 2010. Moraine chronosequence
of the Donnelly Dome region, Alaska. Quat. Res. 74, 63e72. http://
dx.doi.org/10.1016/j.yqres.2010.04.007.
McDowell, P.F., Edwards, M.E., 2001. Evidence of Quaternary climatic variations in a
sequence of loess and related deposits at Birch Creek, Alaska: implications for
the stage 5 climatic chronology. Quat. Sci. Rev. 20, 63e76.
Mock, C.J., Bartlein, P.J., Anderson, P.M., 1998. Atmospheric circulation patterns and
spatial climatic variations in Beringia. Int. J. Climatol. 18, 1085e1104.
Muhs, D.R., Bettis, E.A., 2003. Quaternary loess-paleosol sequences as examples of
climate-driven sedimentary extremes. Geol. Soc. Am. Spec. Pap. 370, 53e74.
Muhs, D., Budahn, J., 2006. Geochemical evidence for the origin of late Quaternary
loess in central Alaska. Can. J. Earth Sci. 43, 323e337.
Muhs, D.R., Ager, T.A., Beann, J.M., Rosenbaum, J.G., Reynolds, R.L., 1998. An evaluation
of methods for identifying and interpreting buried soils in late Quaternary
loess in Alaska. U.S. Geol. Surv. Prof. Pap. 1615, 127e146.
Muhs, D.R., Ager, T.A., Beget, J.E., 2001. Vegetation and paleoclimate of the last
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e2322
interglacial period, central Alaska. Quat. Sci. Rev. 20, 41e61.
Muhs, D.R., Ager, T.A., Bettis III, E.A., McGeehin, J., Been, J.M., Bege
́
t, J.E., Pavich, M.J.,
Stafford Jr., T.W., Stevens, D.A.S.P., 2003. Stratigraphy and palaeoclimatic signiﬁcance
of late Quaternary loess-palaeosol sequences of the Last InterglacialeGlacial
cycle in central Alaska. Quat. Sci. Rev. 22, 1947e1986. http://
dx.doi.org/10.1016/S0277-3791(03)00167-7.
Muhs, D.R., McGeehin, J.P., Beann, J., Fisher, E., 2004. Holocene loess deposition and
soil formation as competing processes, Matanuska Valley, southern Alaska.
Quat. Res. 61, 265e276. http://dx.doi.org/10.1016/j.yqres.2004.02.003.
Muhs, D.R., Ager, T.A., Skipp, G., Beann, J., Budahn, J., McGeehin, J.P., 2008. Paleoclimatic
signiﬁcance of chemical weathering in loess-derived paleosols of
subarctic central Alaska. Arct. Antarct. Alp. Res. 40, 396e411.
Oches, E.A., Banerjee, S.K., Solheid, P.A., Frechen, M., 1998. High- resolution proxies
of climate variability in the Alaskan loess record. In: Busacca, A.J. (Ed.), Dust
Aerosols, Loess Soils and Global Change. Washington State University College of
Agriculture and Home Economics, Miscellaneous Publication No. MISC0190,
Pullman, pp. 167e170.
Pewe, T.L., 1955. Origin of the upland silt near Fairbanks, Alaska. Geol. Soc. Am. Bull.
66, 699e724. http://dx.doi.org/10.1130/0016-7606(1955)66[699:OOTUSN]
2.0.CO;2.
Pewe, T.L., 1975a. Quaternary geology of Alaska. U.S. U.S. Geol. Surv. Prof. Pap. 835,
145.
Pewe, T.L., 1975b. Quaternary stratigraphic nomenclature in central Alaska. U.S.
Geol. Surv. Prof. Pap. 862, 32.
Pewe, T.L., Wahrhaftig, C., Weber, F.R., 1966. Geologic Map of the Fairbanks Quadrangle,
Alaska. U.S. Geol. Surv. Miscellaneous Investigations Map I-455, Scale 1:
250,000.
Pewe, T.L., Berger, G.W., Westgate, J.A., Brown, P.M., Leavitt, S.W., 1997. Eva interglaciation
forest bed, unglaciated East-Central Alaska: global warming 125,000
years ago. Geol. Soc. Am. Spec. Pap. 319, 1e55.
Pewe, T.L., Westgate, J.A., Preece, S.J., Brown, P.M., Leavitt, S.W., 2009. Late Pliocene
Dawson cut Forest bed and new tephrochronological ﬁndings in the gold Hill
loess, east-central Alaska. Geol. Soc. Am. Bull. 121, 294e320. http://dx.doi.org/
10.1130/B26323.1.
Preece, S.J., Westgate, J.A., Stemper, B.A., Pe
́
we
́
, T.L., 1999. Tephrochronology of late
Cenozoic loess at Fair- banks, central Alaska. Geol. Soc. Am. Bull. 111, 71e90.
http://dx.doi.org/10.1130/0016-7606(1999)111<0071:TOLCLA>2.3.CO;2.
Preece, S., Westgate, J., Alloway, B., Milner, M., 2000. Characterization, identity,
distribution, and source of late Cenozoic tephra beds in the Klondike district of
the Yukon, Canada. Can. J. Earth Sci. 37, 983e996.
Preece, S.J., Pearce, N.J.G., Westgate, J.A., Froese, D.G., Jensen, B.J.L., Perkins, W.T.,
2011a. Old Crow tephra across eastern Beringia: a single cataclysmic eruption at
the close of Marine Isotope stage 6. Quat. Sci. Rev. 30, 2069e2090. http://
dx.doi.org/10.1016/j.quascirev.2010.04.020.
Preece, S.J., Westgate, J.A., Froese, D.G., Pearce, N.J.G., Perkins, W.T., 2011b.
A catalogue of late Cenozoic tephra beds in the Klondike goldﬁelds and adjacent
areas, Yukon Territory. Can. J. Earth Sci. 48, 1386e1418.
Reger, R.D., Pinney, D.S., Burk, R.M., Wiltse, M.A., 1996. Catalog and Initial Analyses
of Geologic Data Related to Middle to Late Quaternary Deposits, Cook Inlet
Region, Alaska: State of Alaska Division of Geological and Geophysical Surveys
Report of Investigation 95e96, p. 175. Fairbanks, AK.
Reyes, A.V., Froese, D.G., Jensen, B.J.L., 2010a. Permafrost response to last interglacial
warming: ﬁeld evidence from non-glaciated Yukon and Alaska. Quat. Sci. Rev.
29, 3256e3274. http://dx.doi.org/10.1016/j.quascirev.2010.07.013.
Reyes, A.V., Jensen, B.J.L., Zazula, G.D., Ager, T.A., Kuzmina, S., La Farge, C.,
Froese, D.G., 2010b. A late-Middle Pleistocene (Marine Isotope stage 6) vegetated
surface buried by Old Crow tephra at the Palisades, interior Alaska. Quat.
Sci. Rev. 29, 801e811. http://dx.doi.org/10.1016/j.quascirev.2009.12.003.
Roberts, H.M., 2012. Radiation measurements. Radiat. Meas. 47, 716e724. http://
dx.doi.org/10.1016/j.radmeas.2012.03.022.
Russell, I.C., 1890. Notes on the surface geology of Alaska. Geol. Soc. Am. Bull. 1,
99e162.
Sanborn, P.T., Smith, C.A.S., Froese, D.G., Zazula, G.D., Westgate, J.A., 2006. Fullglacial
paleosols in perennially frozen loess sequences, Klondike goldﬁelds,
Yukon Territory, Canada. Quat. Res. 66, 147e157. http://dx.doi.org/10.1016/
j.yqres.2006.02.008.
Singer, B.S., Hoffman, K.A., 2005. Calibration of a pleistocene geomagnetic instability
time scale (GITS) using 40Ar/39Ar-dated lavas. In: Am. Geophys. Union,
Fall Meeting, Abstract #U42A-02, San Francisco.
Singer, B.S., Guillou, H., Jicha, B.R., Laj, C., Kissel, C., Beard, B.L., Johnson, C.M., 2009.
40
Ar/39
Ar, KeAr and 230
The238
U dating of the Laschamp excursion: a radioisotopic
tie-point for ice core and climate chronologies. Earth Planet. Sci. Lett.
286, 80e88. http://dx.doi.org/10.1016/j.epsl.2009.06.030.
Singer, B.S., 2014. A Quaternary geomagnetic instability time scale. Quat. Geochronol.
21, 29e52. http://dx.doi.org/10.1016/j.quageo.2013.10.003.
Schweger, C.E., 2003. Paleoecology of two marine oxygen isotope stage 7 sites
correlated by the Sheep Creek tephra, northwestern North America. Quat. Res.
60, 44e49. http://dx.doi.org/10.1016/S0033-5894(03)00089-9.
Schweger, C.E., Matthews, J.V., 1985. Early and Middle Wisconsinan environments
of eastern Beringia: stratigraphic and paleoecological implications of the Old
Crow tephra. Geogr. Physique Quaternaire 39, 275e290.
Seidenkrantz, M.S., Bornmalm, L., Johnsen, S.J., Knudsen, K.L., Kuijpers, A.,
Lauritzen, S.E., Leroy, S.A.G., Mergeal, I., Schweger, C., 1996. Two-step deglaciation
at the oxygen isotope stage 6/5e transition: the ZeifeneKattegat climate
oscillation. Quat. Sci. Rev. 15, 63e75.
Spurr, J.E., Goodrich, H.B., 1898. Geology of the Yukon Gold District, Alaska: U.S., vol.
18. Geol. Surv. Annual Rep., pp. 87e392
Taber, S., 1943. Perennially frozen ground in Alaska: its origin and history. Geol. Soc.
Am. Bull. 54, 1433e1548. http://dx.doi.org/10.1130/GSAB-54-1433.
Taber, S., 1953. Origin of Alaska silts. Am. J. Sci. 251, 321e336. http://dx.doi.org/
10.2475/ajs.251.5.321.
Thorson, R.M., Bender, G., 1985. Eolian deﬂation by ancient katabatic winds: a late
Quaternary example from the north Alaska Range. Geol. Soc. Am. Bull. 96,
702e709.
Thouveny, N., Creer, K.M., Blunk, I., 1990. Extension of the Lac du Bouchet palaeomagnetic
record over the last 120,000 years. Earth Planet. Sci. Lett. 97,
140e161.
Thouveny, N., Carcaillet, J., Moreno, E., Leduc, G., Nerini, D., 2004. Geomagnetic
moment variation and paleomagnetic excursions since 400 kyr BP: a stacked
record from sedimentary sequences of the Portuguese margin. Earth Planet. Sci.
Lett. 219, 377e396.
Thouveny, N., Bourles, D., Saracco, G., Carcaillet, J., Bassinot, F., 2008. Paleoclimatic
context of geomagnetic dipole lows and excursions in the Brunhes, clue for an
orbital inﬂuence on the geodynamo? Earth Planet. Sci. Lett. 275, 269e284.
Turner, D., Ward, B.C., Bond, J.D., Jensen, B.J.L., Froese, D.G., Telka, A.M., Zazula, G.D.,
Bigelow, N.H., 2013. Middle to Late Pleistocene ice extents, tephrochronology
and paleoenvironments of the White River area, southwest Yukon. Quat. Sci.
Rev. 75, 59e77. http://dx.doi.org/10.1016/j.quascirev.2013.05.011.
Vlag, P., Oches, E., Banerjee, S., Solheid, P., 1999. The paleoenvironmental-magnetic
record of the Gold Hill Steps loess section in central Alaska. Phys. Chem. Earth
Part A 24, 779e783.
Weber, F., Hamilton, T., Hopkins, D., Repenning, C., Haas, H., 1981. Canyon Creek: a
late Pleistocene vertebrate locality in interior Alaska. Quat. Res. 16, 167e180.
Westgate, J.A., Hamiltion, T.D., Gorton, M.P., 1983. Old Crow tephra: a new late
Pleistocene stratigraphic marker across north-central Alaska and western
Yukon Territory. Quat. Res. 19, 38e54.
Westgate, J., Walter, R., Pearce, G., Gorton, M., 1985. Distribution, stratigraphy,
petrochemistry, and palaeomagnetism of the late Pleistocene Old Crow tephra
in Alaska and the Yukon. Can. J. Earth Sci. 22, 893e906.
Westgate, J.A., Stemper, A.S., Pewe, T.L., 1990. A 3-my record of Pliocene-Pleistocene
loess in interior Alaska. Geology 18, 858e861.
Westgate, J., Preece, S., Kotler, E., Hall, S., 2000. Dawson tephra: a prominent
stratigraphic marker of Late Wisconsinan age in west-central Yukon, Canada.
Can. J. Earth Sci. 37, 621e627.
Westgate, J., Preece, S., Froese, D., Pearce, N., Roberts, R., Demuro, M., Hart, W.,
Perkins, W., 2008. Changing ideas on the identity and stratigraphic signiﬁcance
of the Sheep Creek tephra beds in Alaska and the Yukon Territory, northwestern
North America. Quat. Int. 178, 183e209. http://dx.doi.org/10.1016/
j.quaint.2007.03.009.
Zazula, G.D., Froese, D.G., Westgate, J.A., La Farge, C., Mathewes, R.W., 2005.
Paleoecology of Beringian “packrat” middens from central Yukon Territory,
Canada. Quat. Res. 6, 189e198. http://dx.doi.org/10.1016/j.yqres.2004.11.003.
Zazula, G.D., Froese, D.G., Elias, S.A., Kuzmina, S., Mathewes, R.W., 2007. Arctic
ground squirrels of the mammoth-steppe: paleoecology of late Pleistocene
middens (24,000e29,450 14
C yr BP), Yukon Territory, Canada. Quat. Sci. Rev. 26,
979e1003. http://dx.doi.org/10.1016/j.quascirev.2006.12.006.
Zazula, G.D., Froese, D.G., Elias, S.A., Kuzmina, S., Mathewes, R.W., 2011. Early
Wisconsinan (MIS 4) Arctic ground squirrel middens and a squirrel-eye-view of
the mammoth-steppe. Quat. Sci. Rev. 26 (30), 2220e2237.
B.J.L. Jensen et al. / Quaternary Science Reviews 135 (2016) 1e23 23