Quaternary Research 53, 143–153 (2000)
doi:10.1006/qres.1999.2097, available online at http://www.idealibrary.com on
The Full-Glacial Environment of the Northern Seward Peninsula,
Alaska, Reconstructed from the 21,500-Year-Old Kitluk Paleosol
Claudia Höfle, Mary E. Edwards, 1 David M. Hopkins, and Daniel H. Mann 2
Alaska Quaternary Center, University of Alaska Fairbanks, Fairbanks, Alaska 99775
and
Chien-Lu Ping
Agriculture and Forestry Experiment Station, Palmer Research Center, University of Alaska Fairbanks, 533 E. Fireweed, Palmer, Alaska 99645
Received August 28, 1996
environment of the last glacial maximum (LGM) have focused
further attention on Beringia. The altered geography and fullglacial soil-vegetation system would both have influenced
land-atmosphere feedbacks to the climate system (Kubatzki
and Claussen, 1998; Bartlein et al., 1998). Well-dated empirical information on the full-glacial Beringian land surface is
thus critical for accurate paleoenvironmental and biogeographic reconstruction.
A heterogeneous region such as Beringia was clearly not
covered by a single kind of soil or vegetation. Various authors
have pointed out that landscape-scale differences in slope,
aspect, elevation, and parent material would have maintained a
landscape mosaic (Schweger, 1982; Ritchie and Cwynar, 1982;
Edwards and Armbruster, 1989). Drainages, steep slopes, and
mountain tops differ from low- to mid-elevation interfluves,
which usually support zonal (regionally typical) soils and vegetation. On a larger scale, significant climatic gradients affect
the nature of zonal conditions. Today, for example, tundra
covers maritime central Beringia (western Alaska and
Chukotka), whereas boreal forest dominates the continental
interiors of Alaska–Yukon and Siberia to the east and west.
Climate gradients at the last glacial maximum were likely
different, but still significant. For example, Alaska was probably coldest and driest in the east, close to the Laurentide ice
sheet (Barnosky et al., 1987). Thus, the past environment of
each part of Beringia must be reconstructed from regionally
relevant data, taking both landscape and regional– climatic
conditions into account [e.g., Ritchie and Cwynar (1982) suggest open, discontinuous vegetation of upper slopes, ridges,
and xeric surfaces in the northern Yukon uplands as modern
analogues for zonal full-glacial vegetation of bedrockdominated far-eastern Beringia].
At the last glacial maximum, much of lowland Beringia was
characterized by accumulation of loess (Péwé, 1975; Hopkins,
1982). A widely cited model for the full-glacial environment is
the “arctic steppe” or “mammoth-steppe” hypothesis (Mat-
Paleoenvironmental conditions are reconstructed from soils buried under volcanic ash ca. 21,500 years ago on the Seward Peninsula. Soil development was minimal, reflecting the continuous
regional deposition of loess, which originated from river floodplains and the exposed Chukchi shelf. Cryoturbated soil horizons,
ice wedges, and ice-lens formation indicate a permafrost environment and mean annual temperatures below ⴚ6° to ⴚ8°C. Shallow
active layers (average 45 cm), minimal evidence for chemical
leaching of soils, and the presence of earthen hummocks indicate
a cold and seasonally dry climate. Neither steppe nor polar desert
soils are appropriate analogues for these zonal soils of loesscovered central Beringia. No exact analogues are known; however,
soils underlying dry tundra near the arctic coast of northern
Yakutia, Russia, and under moist, nonacidic tundra of the Alaskan North Slope have properties in common with the buried
soils. © 2000 University of Washington.
Key Words: Beringia; Bering Land Bridge; paleosols;
paleoenvironmental reconstruction; permafrost soils; last glacial
maximum.
INTRODUCTION
As Quaternary sea level fluctuated with glacial cycles, the
shallow Bering and Chukchi shelves intermittently formed a
land connection between Asia and North America: the Bering
land bridge. During the last glacial maximum (ca. 21,500 years
ago), the land bridge lay at the center of a largely ice-free land
mass (Beringia), which stretched between the Kolyma and
Mackenzie rivers (Fig. 1). The Bering land bridge has long
been of interest as a dispersal corridor for plants, animals, and
humans (Hopkins, 1967; Hopkins et al., 1982; West, 1997).
More recently, efforts to reconstruct and model the climate and
1
To whom correspondence should be addressed. Current address: Department of Geography, NTNU (Trondheim), N7034-Dragvoll, Norway.
2
Current address: Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, AK 99775.
143
0033-5894/00 $35.00
Copyright © 2000 by the University of Washington.
All rights of reproduction in any form reserved.
144
HÖFLE ET AL.
FIG. 1. Map showing Beringia (outlined by 100 m isobath) during the last glacial maximum, glacier ice limits, and locations mentioned in the text. The
location of the study area is indicated by the black rectangle. The extent of glacier ice in eastern Beringia is after Mann and Hamilton (1995) and in western
Beringia after Bespalyy (1984).
thews, 1976; Guthrie, 1982, 1990), which is based on diverse
and abundant Pleistocene mammalian fossils, many of which
have been retrieved from loess or loess-derived deposits (e.g.,
Guthrie, 1990; Thorson and Guthrie, 1992). Large grazers such
as mammoth, bison, and horse predominate in these assemblages. Faunal ecology argues for nutritious and relatively
productive vegetation and a firm land surface without a thick
organic mat. Aridity, with little precipitation at any time of
year, and loess deposition would have favored grasses and
tundra forbs over shrubs and would have resulted in higher soil
fertility than today. Guthrie (1990) envisions permafrost because of intense winter cold and thin snow, but an early thaw
and deep active layer because of thin spring snow cover and
strong summer heating. The steppe model is often adopted as
a general description for full-glacial Beringia (e.g., Chapin and
Starfield, 1997). It is important that this and competing models
are tested as rigorously as possible because of their potential
influence on biogeographical and paleoclimatic reconstructions.
Obtaining data from the Bering land bridge is especially
challenging as much of it now lies beneath the sea. Colinvaux
(1964) and Shackleton (1982) cored Imuruk Lake on the
Seward Peninsula, the largest surviving subaerial portion of the
Bering land bridge on the Alaskan side; pollen from the fullglacial sections indicates herb-dominated tundra. Four subsea
peat sections from the Bering Sea dating between 21,000 and
14,000 14C yr B.P. yielded plant and insect fossils indicative of
a relatively mesic tundra with shrub birch (Elias et al., 1996,
1997). At these localities, peat development may have been
associated with topographic lows, and it is hard to estimate
how regionally representative of the land bridge they are.
In this paper we describe the full-glacial soil and land
surface of 10 localities on the northwestern Seward Peninsula.
The aim is to make a paleosol-based reconstruction of what we
assume to be zonal environmental conditions on the upland
surface and to test the predictions of the mammoth-steppe
model for this region. The work complements a detailed paleovegetation description by Wolf and Birks (in press). Detailed pedological descriptions and a discussion of soil genesis
can be found in Höfle and Ping (1996). These reconstructions
form part of the Beringian Natural Heritage Research Program
of the National Park Service (Schaaf, 1995).
FULL-GLACIAL ENVIRONMENT OF SEWARD PENINSULA, ALASKA
STUDY AREA
The study area is located within the boundaries of the Bering
Land Bridge National Preserve on the tundra-covered coastal
plain of northern Seward Peninsula. It is bounded by Kotzebue
Sound on the east and the shore of the Chukchi Sea to the north
(Fig. 2). Relief is low, and altitudes are mostly below 60 m. At
Kotzebue, ca. 60 km to the northeast, January mean temperature is ca. ⫺20°C, July ca. 12°C, and mean annual temperature
(MAT) is ca. ⫺6°C. Precipitation averages 230 mm, with more
than half falling in the summer and early fall (U.S. Weather
Service data). Permafrost is present; measured modern activelayer depths range between 30 and 58 cm under shrub-tussock
tundra on silty substrates and sedge-meadow tundra on peat.
Surface organic layers over mineral soils are 7–25 cm thick.
Further details of modern soils are presented by Höfle et al.
(1998).
During the late Pleistocene, the northern Seward Peninsula
land surface was accreting, due to eolian sediment deposition
and the associated growth of syngenetic ice wedges. Winddeposited loess is almost certainly the parent material, as
indicated by (i) silt–loam textures and grain-size analyses
typical of loess (D. Hopkins, unpublished data), (ii) mineral
composition different from that of underlying basalt in the
nearby Imuruk Lake area (Hopkins, 1963), and (iii) the fact
that the silt drapes the paleolandscape. The loess probably
originated in a now-drowned, braided floodplain that carried
the combined Kobuk and Noatak rivers through the Hope Sea
Valley (Creager and McManus, 1967). Because marine microfossil fragments occur in the loess, the exposed Chukchi sea
bed could also have been a source area (D. Hopkins, unpublished data).
Today, the region experiences active growth of epigenetic
ice wedges. The more massive syngenetic ice wedges form
networks beneath the uneroded late-Pleistocene land surface.
When these first melt, deep thaw lakes are created with surfaces 5–20 m below the surrounding landscape (Carter et al.,
1987; Hopkins and Kidd, 1988; Dinter et al., 1990). In the
course of the Holocene, much of the landscape was worked and
reworked by thermokarst processes, which eventually created
extensive thaw-lake plains with elevations below that of the
late-Pleistocene surface.
About 21,500 cal yr (18,000 14C yr) ago, a maar eruption
created two large craters now occupied by Devil Mountain
Lakes (Fig. 2). The resultant Devil Mountain Lakes tephra
(DMLt) covered about 2500 km 2 of the surrounding area, in
many places to depths of ⬎1 m. The eruption probably continued over a period of weeks or months, with the tephra falling
cold (Begét and Mann, 1992). The plume direction was north
and west, as indicated by the thickest cover of tephra. More
information on tephra distribution and collection localities can
be found in Begét et al. (1996).
With a thick tephra layer insulating the former surface after
the eruption, the frost table must have moved rapidly upward.
145
The soil and vegetation beneath the ash froze, preserving
features of the ancient surface and, where subsequently exposed, providing a unique “snapshot” of the full-glacial landscape. This landscape—low interfluves and shallow valleys
slightly elevated above major floodplains—was probably typical of a considerable portion of full-glacial central Beringia,
and as such probably supported zonal vegetation and soils. We
named the buried soil associated with this event the Kitluk
Paleosol (KP).
The buried surface is now intermittently exposed in actively
eroding bluffs that border thaw lakes (Fig. 2). It is overlain by
DMLt of variable thickness, which in turn is overlain by loess
and a surface peat layer in which the living vegetation is rooted
(Fig. 3). McCulloch and Hopkins (1966) present evidence that
the early Holocene was particularly warm in northwestern
Alaska. If thaw depths in the study area ever exceeded 1–2 m,
it is possible that some of the buried soil horizons previously
lay within the active layer. There is no evidence of slumping,
flowage, or gullying associated with deglacial warming, as
described by Thorson and Guthrie (1992) in interior Alaska.
However, the land surface lacks valleys and steep slopes, and
deeper thaw may have occurred without much landscape destabilization (except lake formation).
No large mammals other than caribou are directly associated
with the paleosurface, although various large-mammal remains, including mammoth, have been recovered from other
late-Pleistocene deposits in northern Seward Peninsula (P.
Heiser, personal communication, 1995). Mammoth remains are
also known from Wisconsin-age sediments on nearby Baldwin
Peninsula (Hopkins et al., 1976).
METHODS
We excavated 10 bluffs during the summers of 1993, 1994,
and 1995, using shovels and a gasoline-powered permafrost
drill to expose the frozen paleosol. In an attempt to confine
observations to soils that have remained frozen since burial, we
chose sites covered with at least 1 m of tephra or, if tephra
coverage was thinner, we required undecayed vegetation to be
in place over the KP, indicating that the active layer had not
reached the paleosol. Table 1 gives details of the localities.
We studied 18 stratigraphic sections. Morphological properties documented were field texture, structure, color, redoximorphic features, and root distribution; the cryogenic features
were ice lenses and ice nets (see below), ice wedges, and
massive ground ice. We analyzed soil samples for particle-size
distribution, pH (H 2O), electrical conductivity, carbonate content, cation exchange capacity and exchangeable cations, active
iron, aluminum, and silica and total carbon, nitrogen, and
sulfur contents. Höfle and Ping (1996) describe 10 soil profiles
in detail; additional information about all profiles can be found
in Höfle (1995).
To confirm that all paleosols were contemporaneous, we
dated organic samples from each location. A suite of 14C
146
HÖFLE ET AL.
FIG. 2.
Location of the study sites on northern Seward Peninsula. This map is based on the 1:250,000 quadrangle Kotzebue.
samples collected in 1992–1994 yielded a large scatter of ages
(Table 2). In 1995 we collected six samples for AMS dating
following a rigorous protocol. We sampled only herbaceous
plant material from the paleosurface, avoiding woody stems,
which might have been decades or centuries old when buried.
Where possible, samples were taken beneath ice, rather than
tephra, as this reduced the possibility of contamination from
above. For each sample, the collector wore new surgical
gloves, and the trowel was rinsed, dipped in alcohol, and
flamed. Samples were placed in new zip-lock bags and immediately sealed.
Ancient active-layer depth was estimated using several cryogenic features, including depths to ice-wedge tops and the
depth of concentrated ice nets and ice lenses. Ice nets are areas
of intersecting horizontal and vertical ice accumulations. Horizontal ice lenses are formed by water moving toward freezing
fronts, and ice accumulates vertically when the soil cracks
during refreezing in the fall (Zhestkova, 1982; Shur, 1988).
Concentrations of ice lenses and ice nets commonly occur in a
zone a few centimeters thick between the active layer and the
permafrost, the “intermediate layer” of Shur and Ping (1994).
The intermediate layer thaws out only during exceptionally
warm summers; its top represents the average base of the active
layer and its base the maximum active-layer depth.
At some sites, earth hummocks cover the paleosurface.
Where these were found, we measured width, length, height,
and distances between hummock centers after removing the
overlying tephra. R. D. Guthrie (Institute of Arctic Biology,
University of Alaska Fairbanks) identified the feces, bones, and
nests of small mammals associated with the paleosols.
RESULTS
Geochronology
The six AMS dates (Table 2) cluster tightly and average
18,070 ⫾ 60 14C yr B.P., or 21,570 cal yr B.P. (using the
CALIB 3.0 program of Stuiver and Reimer, 1993). Thus, the
eruption occurred shortly before 21,500 yr ago (18,000 14C yr),
several hundred years earlier than suggested in previous publications (Goetcheus et al., 1994; Höfle, 1995; Höfle and Ping,
1996; Begét et al., 1996). Two conventionally dated samples
taken farther below the paleosurface date to ca. 19,300 14C yr
B.P.; a third dates to ca. 18,000 14C yr B.P. (Table 2).
Soil Properties
A representative soil profile is shown in Figure 4. Chemical
(Table 3) and physical properties show relatively little variation among profiles and indicate immature soils. All samples
have a silt–loam texture (average 18% sand, 71% silt, 11%
clay), reflecting derivation from loess. Genetic horizons are
poorly developed and difficult to distinguish by morphological
FULL-GLACIAL ENVIRONMENT OF SEWARD PENINSULA, ALASKA
147
FIG. 3. A typical lake bluff section (at Foot Lake, 66°19⬘40⬙N, 164°29⬘50⬙W; described in Höfle, 1995). The upper arrow marks the upper contact of the
tephra and late-Wisconsin loess; the loess is overlain by the Holocene soil and modern vegetation. The middle arrow marks the contact between the tephra and
the underlying paleosurface. The surface has been excavated inward; the arrow points to the center of the resultant shelf. The soil surface is slightly wavy, but
does not show well-developed hummocks. The tape extends 1 m downward. At ca. 50 cm is the top of an ice-rich horizon, which is taken to indicate the average
limit of the active layer. Below 100 cm is slumped material. The soil profile shows mottling and cryoturbation.
properties. Soil horizons are often warped and discontinuous
and are rarely parallel to the surface. Soil colors are predominantly in the 2.5Y and 5Y hue range, with values of 4 or less
and chromas of 2 or less (dark gray, dark grayish brown, and
black). Roots are present throughout all profiles, generally at
high densities (⬎100/dm 2). Most roots are very fine and fine in
148
HÖFLE ET AL.
TABLE 1
Site Localities and Active Layer Depths of Kitluk Paleosol a
Name of site
Lake Rhonda 1
Lake Rhonda 2
Lake Rhonda 3
Ulu Lake
Whitefish Thawpond B
Egg Lake 1
Swan Lake
Tempest Lake A
Tempest Lake B
Fritz Lake
Latitude
(° N)
Longitude
(° W)
Paleoactive-layer
depth (cm)
Estimated by
66°33⬘45⬙
164°27⬘40⬙
66°34⬘20⬙
66°21⬘45⬙
66°33⬘05⬙
66°25⬘50⬙
66°29⬘00⬙
164°23⬘25⬙
164°41⬘00⬙
164°26⬘45⬙
164°41⬘10⬙
164°23⬘00⬙
66°19⬘30⬙
164°21⬘20⬙
50
40
39
52
32
64
45
30
61
40
Ice wedge
Ice wedge
Ice-rich layer
Ice wedge
Ice wedge
Ice net
Ice wedge
Ice net
Ice-rich layer
Ice-rich layer
a
At Whitefish Thawpond B, Swan Lake, Lake Rhonda 1 and 2, and Ulu Lake, maximum depths were determined from ice-wedge tops contemporaneous with
the KP. At Tempest Lake A, Lake Rhonda 3, Fritz Lake, Tempest Lake B, and Egg Lake 1 ice nets and ice-rich horizons gave average depths.
size (⬍1 and 1–2 mm diameter, respectively), although a few
are woody and 2–15 mm in diameter. Soil organic content
ranges from 1.7 to 7.1%. Chemical analyses indicate that (i) the
soil is calcareous, (ii) there is limited chemical weathering of
minerals, (iii) there is no detectable vertical translocation of
ions, and (iv) nutrient status is relatively high. Na values are
high compared with modern soils in the area, which suggests a
marine origin for the loess. Estimates of active-layer depth
TABLE 2
Radiocarbon Dates for Soils and Vegetation Buried beneath DML Tephra
Name of site a
Laboratory
number
Dated material
13
␦ C
(%)
13
C-adjusted
radiocarbon age
( 14C yr B.P.)
1␴ calibrated age
range (cal yr before
A.D. 1950)
⫺28.8
⫺27.9
⫺26.9
⫺28.1
⫺25.2
⫺28.1
⫺25.1
⫺25.0
⫺26.7
⫺27.8
17,420 ⫾ 260
17,850 ⫾ 100
17,980 ⫾ 110
18,140 ⫾ 200
19,630 ⫾ 110
16,880 ⫾ 120
19,990 ⫾ 160
17,740 ⫾ 220
17,360 ⫾ 130
17,880 ⫾ 110 b
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Conventional dates, paleosurface 1992–1994
Lake Rhonda 1
Ulu Lake
Eh’cho Lake
Whitefish Thawpond
Lake Rhonda 3
Egg Lake
Swan Lake
Tempest Lake A ⫹ B
Fritz Lake
Ulu Lake
Beta-60718
Beta-79839
Beta-79837
Beta-79840
Beta-79835
Beta-60716
Beta-79836
Beta-75529
Beta-79838
Beta-81732
Wood from paleosurface
Plant remains from paleosurface
Plant remains from paleosurface
Plant remains from paleosurface
Bulk, uppermost 10 cm of soil
Wood from paleosurface
Bulk, uppermost 10 cm of soil
Wood from paleosurface
Plant remains from paleosurface
Subnivean microtine nest
Conventional dates, below paleosurface 1992–1994
Egg Lake 1
Egg Lake 1
Egg Lake 1
Beta-71982
Beta-71983
Beta-81731
Bulk, buried Ab1 horizon at 30–32 cm depth
Bulk, buried Ab2 horizon at 43–60 cm depth
Nest material, 100 cm depth
⫺27.1
⫺25.1
⫺28.0
17,990 ⫾ 100
19,370 ⫾ 110
19,310 ⫾ 130
Beta-85125
Beta-85126
Beta-85127
Beta-85128
Beta-85129
Beta-85130
Herbaceous
Herbaceous
Herbaceous
Herbaceous
Herbaceous
Herbaceous
⫺27.7
⫺25.3
⫺26.4
⫺27.2
⫺27.2
⫺25.3
18,090 ⫾ 70
17,880 ⫾ 110
17,950 ⫾ 70
18,170 ⫾ 90
18,090 ⫾ 110
18,190 ⫾ 70
AMS dates, paleosurface 1995
Tempest Lake
Whitefish Thawpond
Lake Rhonda 3
Eh’cho Lake
Egg Lake 2
Lake Rhonda 2
a
b
Lake names are informal.
AMS sample taken in 1993.
plant
plant
plant
plant
plant
plant
material
material
material
material
material
material
from
from
from
from
from
from
paleosurface
paleosurface
paleosurface
paleosurface
paleosurface
paleosurface
21,780
21,550
21,610
21,900
21,810
21,900
(21,600)
(21,330)
(21,420)
(21,700)
(21,600)
(21,725)
21,400
21,100
21,230
21,500
21,380
21,540
149
FULL-GLACIAL ENVIRONMENT OF SEWARD PENINSULA, ALASKA
(Microtus miurus). Similar feces, ca. 1 ⫻ 3 mm in size, were
scattered throughout many profiles.
DISCUSSION
Physical Environment
FIG. 4. A typical Kitluk Paleosol profile from Swan Lake, showing an
upper weathered B and a lower, unweathered B horizon, with mottles distributed throughout most of the profile, especially at depth. The top of an ice
wedge marks the active-layer depth at 45 cm.
beneath the KP range from 30 to 64 cm and average 45 cm
(Table 1). At the time of sampling, the buried soil horizons
commonly were saturated, or oversaturated, with water present
as fine lenses of segregation ice. If deep thawing during the
warmest part of the Holocene affected some buried soil horizons, it may explain this excess water, but, as mentioned
above, there is no clear evidence that this took place at the
sections we selected for study.
Microrelief, Ice at the Buried Surface, and Faunal Remains
Of the 18 exposures, the top surfaces of 7 are hummocky, 3
are slightly wavy (relief of only few centimeters), and 8 have
a flat surface. Hummock dimensions (range and mean, in
centimeters) are 10 –32 (21) wide, 19 – 40 (29) long, and 9 –20
(13) high. Center-to-center distances are 18 –55 (35). At several sites in low places on the paleosurface, we found flat-top
masses of clear ice encasing undamaged plants in growth
position. Some of the ice had tephra fragments embedded
in it. We also found masses of slightly opaque to clear ice with
uneven upper surfaces.
A small mammal nest and adjacent void at ca. 100 cm depth
at Egg Lake 1 contained woody and herbaceous plant material,
the mandible of a collared lemming (Dicrostonyx torquatus),
feces of D. torquatus, and, possibly, feces of the singing vole
With continual loess input, the KP surface was presumably
continually renewed. If the 19,370 ⫾ 110 14C yr B.P. date for
the Ab2 horizon (43– 60 cm) at Egg Lake 1 is accurate (Table
2), and if the permafrost table rose in synchrony with the
deposition of loess (ca. 0.5 mm/yr), new material turned into
permafrost within a few thousand years of deposition, preventing soil development beyond an incipient stage, as indicated by
the KP chemical and morphological properties. Rates are similar to those estimated for loess accumulation in middle Wisconsin time in the Fairbanks area (0.75– 0.375 mm/yr: Hamilton et al., 1988).
Circumstantial evidence suggests that the eruption occurred
during the spring. We interpret the clear, flat-top ice with intact
plant material as frozen puddles and the opaque ice with an
uneven surface as snow. The snow was evidently patchy,
suggesting a landscape during break-up. Based on modern
analogues, the puddles would have formed from runoff that
froze nightly. The tephra fragments suggest that puddle ice
broke under the first tephra fall or that tephra fell into partly
thawed puddles, which then refroze. Goetcheus et al. (1994)
report an intact nest of M. miurus on the paleosurface, which
also suggests winter or spring, as nests become dispersed
during the summer. In this scenario, saturation of the KP
reflects conditions in fall, immediately prior to freeze-up (see
below). It seems unlikely (but not impossible) that the soils
became saturated by water percolating from above after burial,
given the presence of patchy ice on the paleosurface, rather
than a continuous ice layer, and the gray soil colors that
indicate gleying as an important soil-forming process.
Earthen hummocks are a form of patterned ground, but their
formative processes remain poorly understood (Mackay,
1980). Hummocks of similar dimensions to those we observed
are reported from polar and subpolar deserts (Tedrow, 1977).
D. Hopkins (unpublished data) also noted them in interior
TABLE 3
Summary of Chemical Properties of Kitluk Paleosol
Total
Range:
Minimum
Maximum
Mean value
Standard deviation
CE
Exchangeable cations
(cmol/kg)
Amm. oxalate
(%)
Pyrophosphate
(%)
Dith.-Citr.
(%)
pH
(paste)
EC
(S/m)
OC
(%)
N
(%)
S
(%)
C/N
(%)
CO 3
(%)
C
(%)
Ca
Mg
K
Na
Fe
Al
Si
Fe
Al
Fe
Al
7.4
8.6
8.0
0.2
0.06
1.00
0.38
0.30
1.7
7.1
3.0
1.0
0.19
0.55
0.32
0.10
0.04
0.70
0.10
0.10
8
13
9
1.1
0.5
4.3
2.2
0.9
14.3
32.0
19.1
3.6
16.7
52.1
31.0
6.1
4.4
15.9
7.9
2.8
0.1
0.5
0.3
0.1
0.2
8.3
2.9
2.4
0.87
4.74
1.96
0.80
0.16
1.06
0.44
0.20
0.00
0.81
0.23
0.20
0.24
1.30
0.52
0.30
0.00
0.33
0.10
0.10
0.62
4.33
1.78
0.70
0.08
0.35
0.17
0.10
150
HÖFLE ET AL.
Alaska, where they were developed in 30 – 40 cm of seasonally
dry loess overlying shattered bedrock on a south-facing slope
of about 20°. These seemed to be stable, despite the steep
slope, and to be related to the interaction of long-term wetting
and drying cycles and the positions of plant root masses.
Possibly the KP hummocks, which are associated with shallow
slopes on the paleolandscape, were initiated by drying and
cracking of the soil surface in summer, followed by saturation
in late summer and fall and then growth of ice lenses and soil
expansion during freeze-up. Whatever their origin, they evidently developed into stable perennial features, as mosses were
frequently observed in growth position in interhummock spaces.
Biotic Environment
The paleovegetation of the KP surface, as described by Wolf
and Birks (in press), is characterized by graminoids and forbs,
with arctic willow confined to certain areas, probably snowbeds. The sedge Kobresia myosuroides is dominant, and the
paleovegetation exhibits a fine-scale compositional mosaic related to moisture availability. The vegetation is interpreted as a
dry tundra or meadow tundra, probably closed, with a continuous acrocarpous moss layer. Wolf and Birks (in press) suggest
that the best physiognomic modern analogues for this vegetation are the dry meadow and Kobresia meadow communities of
Yurtsev (1982), although species composition differs. D.
Walker (personal communication, 1997) has pointed out the
physiognomic similarities between moist, nonacidic tundra of
the Alaskan North Slope (Walker et al., 1998) and the paleovegetation, although again, species composition is different.
Collared lemmings are browsers; today they eat primarily
willow leaves and Dryas (the latter being absent from the
sampled paleovegetation). They are commonly found in dry,
open-ground environments. Singing voles thrive best in discontinuous, drifted-snow conditions, with a minimum snow
cover of 20 –30 cm (D. Guthrie, personal communication, 1994).
Paleoclimatic Inferences
The KP developed in a cold, permafrost environment. This
is clearly indicated by the presence of ice wedges underlying
the soil, platy and blocky structures in the soil reflecting the
formation of ice lenses and ice nets, undecayed fine roots even
in the lowest parts of the profiles, and discontinuous and
warped soil horizons that are typical for soils formed under the
influence of cryoturbation (Bockheim et al., 1994). A mean
annual air temperature at least as cold as ⫺6° to ⫺8°C is
indicated by the growth of ice wedges (Washburn, 1980).
Snow cover was probably thin, but with drifts of at least 20 –30
cm, as indicated by the mammalian remains.
KP chemical properties indicate a dry, but not arid, soilforming environment. In a continually wet, leaching environment carbonates and other soluble salts would be leached out
of the soil profile or accumulate at depth. Although the apparently saturated status of the KP and its gray color indicate at
least seasonal wetness, high exchangeable Na values show no
changes with depth, and carbonates have not been altered or
transported by pedogenic processes. The fact that loess was a
major environmental feature itself argues for at least seasonal
dryness. In an extremely arid environment soluble salts would
be moved upward by surface evaporation, a common process
in modern polar desert soils (Tedrow, 1977), but this is not
evident in the KP.
Active-layer depths respond to temperature, precipitation,
and seasonality patterns, as well as vegetation, thickness of
organic mat, slope, aspect, and drainage. If we assume topography to have been similar in the LGM, the main nonclimate
feature that was different from modern times is the lack of a
surface organic layer, which should lead to deeper thaw. In
contrast, much colder MATs and/or seasonal soil wetness
would tend to reduce active-layer depth. Soils that are saturated
going into winter require more energy than drier soils to thaw
the following spring, which reduces eventual active-layer
thickness (V. Romanovsky, personal communication, 1995).
This scenario would fit with the saturated condition of the
buried soils and would indicate a strongly seasonal precipitation, with most occurring in late summer and fall. Strong
heating during a dry summer could still evaporate enough
water to dry the soil surface, initiate hummock development,
and allow root penetration throughout the active layer. This
model could apply to the KP, if we are correct in inferring that
the saturation reflects preburial conditions.
On the other hand, if the presently observed saturation is a
postburial feature, the KP may have been relatively dry all
year. In this case, a considerably colder MAT than present
would have been required to produce active-layer depths similar to modern but developed in the absence of an insulating
organic mat. There are few other data on active layer depths at
the LGM. Thorson and Guthrie (1992) estimate that in loessaffected interior Alaska the active layer may have been as little
as 20 cm deep, which implies very low MATs, but it is not
known whether their one location is typical. In the future we
may be able to develop paleoclimatic reconstructions further
via numerical modeling of active-layer dynamics.
Modern Analogues for the KP
Surface dryness, abundant fine roots, and lack of a surface
peat horizon are broadly consistent with both steppe and dry
herbaceous tundra soils. Although unusual, permafrost has
been documented under steppe in Yakutia, northeastern Russia,
and on the Qinghai-Xizang (Tibetan) plateau, China (C-L.
Ping, personal observation, 1999). In these cases, active-layer
depth is deeper than that of the KP (ca. 1.0 –2.0 m). We know of
two modern tundra–soil descriptions somewhat similar to the KP.
One is from loess-covered, gently rolling interfluves near the
Arctic coast of northern Yakutia, Russia, observed by C-L.
Ping and S. Gubin in 1992. The soils have a poorly developed
organic horizon [0 –5 (10) cm], a grey A horizon (5–10 cm), a
greyish– brown B horizon (10 –15 cm), and a bluish– grey Bg
FULL-GLACIAL ENVIRONMENT OF SEWARD PENINSULA, ALASKA
horizon, (15–20 cm). The active layer is ca. 45 cm deep. Frost
boils and active surface cracks were observed, and at least a
third of the profiles were undergoing cryoturbation. The pH of
the B horizon was 6.5–7.5. The associated vegetation is dominated by Dryas, dwarf willow, herbs, and graminoids.
The second possible analogue is soils under the moist nonacidic tundra mentioned above (Walker et al., 1998). These are
characterized by high base saturation and pH, a gleyed C
horizon, and cryoturbation. Measured thaw depths are 52 to 57
cm. In contrast to the KP, they have a dark, organic-rich A
horizon. Although no long-term climate means are available
for the measured MNT soils, means for the two nearest stations, Umiat and Arctic Village, are 12.4 and 14.2°C (July),
⫺29.4 and ⫺30.5°C (January), and 127 and 223 mm (mean
annual precipitation), respectively (Western Regional Climate
Center data). The MNT is thus characterized by similar summer temperatures, but colder winters and probably lower annual precipitation than is the KP study area today.
Another feature differentiating the KP from most modern
soils is the steady influx of loess. At Kluane Lake, southwestern Yukon Territory, modern loess deposition occurs on local
areas of grass-dominated vegetation. Both plant biomass and
species diversity increase with increasing dust influx (Laxton et
al., 1996). The KP vegetation would have been subject to this
“fertilization enhancement” of diversity and productivity (Wolf
and Birks, in press).
We can hypothesize a soil–vegetation continuum under generally cool to cold, dry climates, even though parts may not be
represented by widespread modern analogues, as much of the
modern Arctic is relatively moist. Major gradients would involve nutrient availability and soil temperature, both of which
would tend to vary reciprocally with effective moisture. Locally, slope and aspect would be a major determinant of soil
temperature, through their effect on insolation receipt and
drainage (Edwards and Armbruster, 1989). Regional gradients
in temperature and precipitation would have an effect at larger
scales. At one extreme would be sparse, dry tundra, perhaps
akin to polar semi-desert on extremely dry soils (Bliss and
Richards, 1982). At the other would be productive, grassdominated vegetation on warm, permafrost-free soils, such as
that described by Laxton et al. (1996), which could today be
described as azonal steppe. Based on KP data and the partial
analogues, we suggest that the KP and its associated vegetation
lies somewhere in the middle of this continuum and that it
represents a cooler, shorter, and drier growing season than that
of the region today. Slow drying that may have kept summer
soil temperatures relatively low would have been countered by
the high nutrient status due to loess accumulation.
Testing the “Mammoth Steppe” Hypothesis
Guthrie’s (1990) model provides testable predictions. In
accord with the model, the KP surface organic horizon was thin
or absent. Although the KP surface must have been seasonally
soft, it was not a quagmire. The hummocks maintained their
151
shape, and the surface was probably firm most of the year, as
required by the relatively small-hooved megafauna. The dominance and relatively dense cover of graminoids and forbs are
also in accord with Guthrie’s predictions. However, the shallow active-layer depths in the KP are counter to the model. A
concentration of fall precipitation may explain the failure of the
active-layer prediction, which does not consider the retarded
thawing of seasonally saturated soils. The retarded thaw model
implies lower productivity than that envisioned by Guthrie
(1990), but nevertheless the KP landscape may have been as, or
more, productive than many modern tundra landscapes with
respect to consumable plant material.
The results of this study suggest that in many details the soil
attributes of the ecosystem described by Guthrie (1990) are
close to those reconstructed for the KP surface. However, a
major drawback of the mammoth-steppe as a model for a
widespread, loess-dominated, Beringian ecosystem in fullglacial time relates to nomenclature. The image of typical
steppe that tends to be invoked is misleading. The KP soil is
not a typical steppe soil, nor do Wolf and Birks (in press)
interpret the vegetation as steppe vegetation. The paleoclimatic
inferences from the KP data suggest that the soil formed under
climatic conditions that today would be associated with tundra.
ACKNOWLEDGMENTS
We thank the U.S. National Park Service’s Beringian Natural Heritage
Program and park service personnel in Kotzebue and Nome for financial and
logistical support. We especially thank Jeanne Schaaf and Dale Taylor (NPS)
for their encouragement and enthusiasm throughout the project. We are also
grateful for financial assistance from the USDA Global Change Initiative
Program. Jim Begét, Scott Elias, Vicky Goetcheus, Dale Guthrie, Dietrich
Lemme, Vladimir Romanovsky, Yuri Shur, Scott Smith, and Dave Swanson
engaged in useful discussions about the data. Field help was provided by
Tiffany Fraser, Larry Huber, John Kimble, Dietrich Lemme, Lawrence Plug,
Yuri Shur, and Rhonda Weyiowanna. R. Thorson and C. Waythomas provided
helpful reviews of an early draft of the manuscript.
REFERENCES
Barnosky, C. W., Anderson, P. M., and Bartlein, P. J. (1987). The northwestern
U.S. during deglaciation; Vegetational history and paleoclimatic implications. In “North America and adjacent oceans during the past deglaciation,
The geology of North America, Vol. K-3” (W. F. Ruddiman and H. E.
Wright, Jr., Eds.), pp. 289 –321. Geological Society of America, Boulder,
CO.
Bartlein, P. J., Anderson, K. H., Anderson, P. M., Edwards, M. E., Mock, C. J.,
Thompson, R. S., Webb, R. S., Webb, T., III, and Whitlock, C. (1998).
Paleoclimate simulations for North America over the past 21,000 years:
Features of the simulated climate and comparisons with paleoenvironmental
data. Quaternary Science Reviews 17, 549 –585.
Begét, J., Hopkins, D. M., and Charron, S. D. (1996). The largest known maars
on earth, Seward Peninsula, northwest Alaska. Arctic 49(1), 62– 69.
Begét, J., and Mann, D. H. (1992). “Caldera” formation by unusually large
phreatomagmatic eruptions through permafrost in arctic Alaska. Eos 73,
636.
Bespalyy, V. G. (1984). Late Pleistocene mountain glaciation of the northeast-
152
HÖFLE ET AL.
ern USSR. In “Late Quaternary Environments of the Soviet Union” (A. A.
Velichko, Ed.), pp. 31–33. University of Minnesota Press, Minneapolis.
Hopkins, D. M. (1963). Geology of the Imuruk Lake area, Seward Peninsula,
Alaska: U. S. Geological Survey Bulletin 1141-c, 100 pp.
Bliss, L. C., and Richards, J. H. (1982). Present-day arctic vegetation and ecosystems as a predictive tool for the arctic-steppe mammoth biome. In “Paleoecology of Beringia” (D. M. Hopkins, J. V. Matthews, C. E. Schweger, and S. B.
Young, Eds.), pp. 241–257. Academic Press, New York.
Hopkins, D. M. (1982). Aspects of the paleogeography of Beringia during the
late Pleistocene. In “Paleoecology of Beringia” (D. M. Hopkins, J. V.
Matthews, C. E. Schweger, and S. B. Young, Eds.), pp. 3–28. Academic
Press, New York.
Bockheim, J. G., Ping, C. L., Moore, J. P., and Kimble, J. M. (1994). Gelisols:
A new proposed order for permafrost-affected soils. In “Proceedings of the
Meeting of the Classification, Correlation, and Management of PermafrostAffected Soils, July, 1993” (J. M. Kimble and R. J. Ahrens, Eds.), pp.
25– 44. USDA, Soil Conservation Service, National Soil Survey Center,
Lincoln, NE.
Hopkins, D. M., and Kidd, J. G. (1988). Thaw-lake sediments and sedimentary
environments. In “Fifth International Permafrost Conference Proceedings,”
Vol. 1, pp. 790 –795. Trondheim, Norway.
Carter, L. D., Heninbottom, J. A., and Woo, M. K. (1987). Arctic lowlands. In
“Geomorphic Systems of North America” (W. L. Graf., Ed.), pp. 583– 628.
Geol. Soc. Amer. Centennial Special Vol. 2, Boulder, CO.
Hopkins, D. M., Matthews, J. V., Schweger, C. E., and Young, S. B. (Eds.)
(1982). “Paleoecology of Beringia.” Academic Press, New York.
Chapin, F. S., and Starfield, A. M. (1997). Time lags and novel ecosystems in
response to transient climatic change in arctic Alaska. Climatic Change 35,
449 – 461.
Hopkins, D. M., Giterman, R. E., and Matthews, J. V., Jr. (1976). Interstadial
mammoth remains and associated pollen and insect fossils, Kotzebue Sound
area, northwestern Alaska. Geology 4, 169 –172.
Kubatzki, C., and Claussen, M. (1998). Simulation of the global biogeophysical interactions during the Last Glacial Maximum. Climate Dynamics 14, 461– 471.
Colinvaux, P. A. (1964). The environment of the Bering Land Bridge. Ecological Monographs 34, 297–329.
Laxton, N. F., Burn, C. R., and Smith, C. A. S. (1996). Productivity of loessal
grasslands in the Kluane Lake region, Yukon Territory, and the Beringian
“production paradox.” Arctic 49, 129 –140.
Creager, J. S., and McManus, D. A. (1967). Geology of the floor of Bering and
Chukchi Seas—American studies. In “The Bering Land Bridge” (D. M.
Hopkins, Ed.), pp. 7–31. Stanford Univ. Press, Stanford.
Mackay, J. R. (1980). The origin of hummocks, western Arctic Coast, Canada.
Canadian Journal of Earth Sciences 17, 996 –1006.
Dinter, D. A., Carter, L. D., and Brigham-Grette, J. (1990). Late Cenozoic
geologic evolution of the Alaskan North Slope and adjacent continental
shelves. In “The Arctic Ocean Region, The Geology of North America, Vol.
L,” (A. Grantz, L. Johnson, and J. F. Sweeney, Eds.), pp. 459 – 490.
Geological Society of America, Boulder, CO.
Mann, D. H., and Hamilton, T. H. (1995). Late-Pleistocene and Holocene
paleoenvironments of the North Pacific coast. Quaternary Science Reviews
14, 449 – 471.
Matthews, J. V., Jr. (1976). Arctic steppe—An extinct biome. AMQUA Abstracts 4, 73–77.
Edwards, M. E., and Armbruster, W. S. (1989). A tundra-steppe transition on
Kathul Mountain, Alaska, USA. Arctic and Alpine Research, 21(3), 296 –304.
McCulloch, D., and Hopkins, D. M. (1966). Evidence for an early Recent
warm interval in northwestern Alaska. Geological Society of America Bulletin 77, 1089 –1108.
Elias, S. A., Short, S. K., Nelson, C. H., and Birks, H. H. (1996). Life and times
of the Bering Land Bridge. Nature 382, 60 – 63.
Péwé, T. L. (1975). “Quaternary Geology of Alaska.” Geological Survey Paper
835. U.S. Government Printing Office, Washington, DC.
Elias, S. A., Short, S. K., and Birks, H. H. (1997). Late Wisconsin environments of the Bering Land Bridge. Palaeogeography, Palaeoclimatology,
Palaeoecology 136, 293–308.
Ritchie, J. C., and Cwynar, L. C. (1982). The late-Quaternary vegetation of the
Northern Yukon. In “Paleoecology of Beringia” (D. M. Hopkins, J. V.
Matthews, C. E. Schweger, and S. B. Young, Eds.), pp. 113–126. Academic
Press, New York.
Goetcheus, V. G., Hopkins, D. M., Edwards, M. E., and Mann, D. H. (1994).
Window on the Bering Land Bridge: A 17,000-year-old paleosurface on the
Seward Peninsula, Alaska. Current Research in the Pleistocene 11, 131–
132.
Guthrie, R. D. (1982). Mammals of the mammoth-steppe as paleoenvironmental indicators. In “Paleoecology of Beringia” (D. M. Hopkins, J. V. Matthews, C. E. Schweger, and S. B. Young, Eds.), pp. 307–326. Academic
Press, New York.
Guthrie, R. D. (1990). “Frozen fauna of the mammoth steppe. The story of
Blue Babe.” Univ. of Chicago Press, Chicago.
Hamilton, T. D., Craig, J. L., and Sellmann, P. V. (1988). The Fox permaforst
tunnel: A late Quaternary geologic record in central Alaska. Geological
Society of America Bulletin 100, 948 –969.
Höfle, C. (1995). Buried soils on Seward Peninsula, northwest Alaska: A
window into the late-Pleistocene environment of the Bering Land Bridge.
M.Sc. Thesis, University of Alaska–Fairbanks. Fairbanks, AK.
Höfle, C., and Ping, C. L. (1996). Properties and soil development of latePleistocene paleosols from Seward Peninsula, northwest Alaska. Geoderma
71, 219 –243.
Höfle, C., Ping, C.-L., and Kimble, J. M. (1998). Properties of permafrost soils
on Northern Seward Peninsula, Northwest Alaska. Soil Science Society of
America Journal 62, 1629 –1639.
Hopkins, D. M. (Ed.) (1967). “The Bering Land Bridge.” Stanford Univ. Press,
Stanford.
Schaaf, J. M. (1995). Understanding northern environments and human populations through cooperative research: a case study in Beringia. In “Human
Ecology and Climate Change: People and Resources in the Far North” (D. L.
Peterson and D. R. Johnson, Eds.) Taylor and Francis, Washington, DC.
Schweger, C. E. (1982). Late Pleistocene vegetation of eastern Beringia:
Pollen analysis of date alluvium. In “Paleoecology of Beringia” (D. M.
Hopkins, J. V. Matthews, C. E. Schweger, and S. B. Young, Eds.), pp.
95–112. Academic Press, New York.
Shackleton, J. (1982). “Environmental histories from Whitefish and Imuruk
Lakes, Seward Peninsula, Alaska”. Institute of Polar Studies Report No. 76.
Ohio State Univ., Columbus. 49 pp.
Shur, Y. L. (1988). “Verkhniy Gorizont Tolshchi Merzlykh Porod I Termokarst. (Thermokarst and the upper horizon of permafrost.)” Nauka, Moscow, USSR. (In Russian).
Shur, Y. L., and Ping, C. L. (1994). Permafrost dynamics and soil formation.
In “Proceedings of the Meeting of the Classification, Correlation, and
Management of Permafrost-Affected Soils, July, 1993” (J. M. Kimble and
R. J. Ahrens, Eds.), pp. 112–117. USDA, Soil Conservation Service, National Soil Survey Center, Lincoln, NE.
Stuiver, M., and Reimer, P. J. (1993). Extended 14C data base and revised
CALIB 3.0 14C age calibration program. Radiocarbon 35, 215–230.
Tedrow, J. C. F. (1977). “Soils of the polar landscapes.” Rutgers Univ. Press,
New Brunswick, NJ.
FULL-GLACIAL ENVIRONMENT OF SEWARD PENINSULA, ALASKA
Thorson, R. M., and Guthrie, R. D. (1992). Stratigraphy of the Colorado Creek
mammoth locality, Alaska. Quaternary Research 37, 214 –228.
Walker, D. A., Auerbach, N. A., Bockheim, J. G., Chapin, F. S. III, Eugster,
W., Kling, J. Y., McFadden, J. P., Michaelson, G. J., Nelson, F. E., Oechel,
W. C., Ping, C. L., Reeburg, W. S., Regli, S., Shiklomanov, N. I., and
Vourlitis, G. L. (1998). Energy and trace-gas fluxes across a soil pH
boundary in the Arctic. Nature 394, 469 – 472.
Washburn, A. L. (1980). “Geocryology.” Wiley, New York.
West, F. H. (Ed.) (1997). “American Beginnings.” Univ. of Chicago Press,
Chicago.
153
Wolf, V. G., and Birks, H. H. (In press). Full-glacial upland tundra vegetation
preserved under tephra in the Beringia National Park, Seward Peninsula.
Quaternary Science Reviews.
Yurtsev, B. A. (1982). Relics of the xerophyte vegetation of Beringia in
northeastern Asia. In “Paleoecology of Beringia” (D. M. Hopkins, J. V.
Matthews, C. E. Schweger and S. B. Young, Eds.). pp. 157–177. Academic
Press, New York.
Zhestkova, T. N. (1982). “Formirovaniye Kriogennogo Stroeniya Gruntov.
(The formation of cryogenic soil structure).” Siberian Branch of Academy of
Science, USSR. Permafrost Institute. Nauka, Moscow, USSR (in Russian).