336 VOLUME 14JOURNAL OF CLIMATE
q 2001 American Meteorological Society
Snow–Shrub Interactions in Arctic Tundra: A Hypothesis with Climatic Implications
MATTHEW STURM,* JOSEPH P. M CFADDEN,
1,
** GLEN E. LISTON,
#
F. S TUART CHAPIN III,
@
CHARLES H. RACINE,
&
AND JON HOLMGREN*
* U.S. Army Cold Regions Research and Engineering Laboratory, Fort Wainwright, Alaska
1
Department of Integrative Biology, University of California, Berkeley, Berkeley, California
# Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado
@
Institute of Arctic Biology, University of Alaska, Fairbanks, Fairbanks, Alaska
&
U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire
(Manuscript received 3 January 2000, in final form 14 March 2000)
ABSTRACT
In the Arctic, where wind transport of snow is common, the depth and insulative properties of the snow cover
can be determined as much by the wind as by spatial variations in precipitation. Where shrubs are more abundant
and larger, greater amounts of drifting snow are trapped and suffer less loss due to sublimation. The snow in
shrub patches is both thicker and a better thermal insulator per unit thickness than the snow outside of shrub
patches. As a consequence, winter soil surface temperatures are substantially higher, a condition that can promote
greater winter decomposition and nutrient release, thereby providing a positive feedback that could enhance
shrub growth. If the abundance, size, and coverage of arctic shrubs increases in response to climate warming,
as is expected, snow–shrub interactions could cause a widespread increase (estimated 10%–25%) in the winter
snow depth. This would increase spring runoff, winter soil temperatures, and probably winter CO
2
emissions.
The balance between these winter effects and changes in the summer energy balance associated with the increase
in shrubs probably depends on shrub density, with the threshold for winter snow trapping occurring at lower
densities than the threshold for summer effects such as shading. It is suggested that snow–shrub interactions
warrant further investigation as a possible factor contributing to the transition of the arctic land surface from
moist graminoid tundra to shrub tundra in response to climatic warming.
1. Introduction
Air temperatures in Alaska and other parts of the
Arctic have increased (Chapman and Walsh 1993), and
climate simulations suggest that any continued warming
will be greater in the Arctic than in lower latitudes (Kat-
tenberg et al. 1996). One expected result of warming is
an increase in plant productivity, which may be reflected
in recent increases in the seasonal amplitude of high-
latitude CO
2
(Keeling et al. 1996; Zimov et al. 1999)
and in the seasonally integrated normalized difference
vegetation index (an index of plant productivity) (My-
neni et al. 1997), although other interpretations of these
data are possible.
If the productivity of the tundra rises, an increase in
the height and abundance of shrubs is likely to be one
important outcome. Transects along climatic gradients
** Current affiliation: Department of Atmospheric Science, Col-
orado State University, Fort Collins, Colorado.
Corresponding author address: Dr. Matthew Sturm, USA-CRREL-
Alaska, P.O. Box 35170, Fort Wainwright, AK 99703-0170.
E-mail: msturm@crrel.usace.army.mil.
show that shrub tundra replaces tussock tundra near the
southern tundra limit where the climate is warmer (Alek-
sandrova 1980; Bliss and Matveyeva 1992). Similarly,
Holocene warming was accompanied by expansion of
Betula (birch) and other shrubs (Ager 1983; Payette et
al. 1989; Anderson and Brubaker 1993; Brubaker et al.
1995; Jacoby and D’Arrigo 1995). Some manipulation
experiments in which growing season temperatures were
elevated have also resulted in an increase in shrubs
(Hobbie and Chapin 1998), though other similar ex-
periments have not shown the same temperature effect
(e.g., Chapin et al. 1995).
An increase in shrub abundance would have impor-
tant implications for regional climate in the Arctic. In
summer, changes in energy partitioning between the
shrub canopy and the ground could lead to changes in
shading and active layer thickness. In winter, shrubs and
snow would interact in several ways. Because the Arctic
is windy and snow-covered 9 months of the year, snow
drifted by the wind is trapped by shrubs. In this paper
we show that an increase in shrubs could augment the
depth of snow on the ground, both locally and generally,
in part by diminishing winter water losses caused by
wind-driven sublimation. We also show that when the
snow depth in and around shrubs is increased, higher

1FEBRUARY 2001 337STURM ET AL.
FIG. 1. Spatial variations in snow depth, canopy height, and topographic relief along transects (a) from shrub to tussock tundra and (b)
from shrubby tussock to tussock tundra. Note that near 700 m in the first transect there is an area of deeper snow and taller canopies not
associated with a topographic depression. The D elevation was computed by subtracting a 50-m moving average of elevation from the
measured elevation, producing a profile that highlights the local relief.
subnivian temperatures result. We suggest that at these
higher temperatures, more winter decomposition and
nutrient mineralization may occur, producing more fa-
vorable conditions for the growth of shrubs. In this paper
we point out the existence and the potential importance
of these winter biogeophysical linkages, and suggest
that they play a role in the general response of the tundra
to climate change.
2. Study design
We measured variations in snow properties and veg-
etation across a landscape in arctic Alaska (698069N,
1498009W) covered by three types of vegetation: 1) tus-
sock tundra, 2) shrubby tussock tundra, and 3) riparian
shrub (McFadden et al. 1998). The site was near Happy
Valley on the Dalton Highway, about half-way between
Prudhoe Bay and the Brooks Range. Several shallow
water tracks drained the gently sloping area. These were
oriented perpendicular to the prevailing winter wind and
filled with drifting snow. Outside of water tracks, a thin
(average 0.6 m), wind-blown snow cover developed at
the site (Benson and Sturm 1993). In April 1996, when
the snow cover had reached maximum depth, we mea-
sured an extensive set of snow properties (depth, den-
sity, stratigraphy, thermal conductivity, and the tem-
perature of the snow–ground interface) along intersect-
ing traverse lines through the three vegetation types,
each line being about 1 km long. In July we returned
and recorded the topography, vegetation species, canopy
height, stem thickness, and leaf area index (LAI) along
the same lines. Canopy height was taken as the average
height of the five tallest shrubs at each measurement
station, and stem thickness was a similar average of the
diameter of five randomly chosen stems, measured at
the base of the plant using a caliper. LAI, leaf area per
unit ground area, was measured using an optical plant
canopy analyzer (LI-COR, Inc., LAI-2000). Detailed
snow and vegetation measurements (data every meter
for 200 m) were measured at the intersections of the
traverse lines in the three vegetation types. Both snow
and vegetation measurements were extrapolated spa-
tially using aerial photographs taken in May (partial
snow cover) and August (no snow). Continuous snow–
ground interface temperature records (60.78C) were
collected nearby using mini-dataloggers (see http://
arcss.colorado.edu/Catalog/arcss001.html).
3. Results
The deepest snow was associated with the tallest,
densest shrubs (Figs. 1a and 1b). These were often near
water tracks or in riparian areas. Some, but not all, of
the increase in snow depth associated with the tall shrubs
was the result of in-filling of water track channels. How-
ever, the topographic depressions made by the water
tracks were quite shallow and no more than 10 m wide,
while the deeper snow associated with the tracks was
50–60 m wide (Fig. 1a). This, along with a lag corre-
lation analysis between shrub height, snow depth, and
local relief (McFadden 1998), established that the shrub
canopy, rather than the topographic relief, was the main
control on the snow.
There was about a 10-m downwind displacement of
the deepest snow from the tallest shrubs (Fig. 2, top
panels). Shrub height, stem diameter, and leaf area all
declined more quickly downwind than did the snow
depth, and these downwind changes roughly corre-
sponded to the transition from Salix (willow)-dominated
riparian areas to Betula (birch)-dominated shrubby-tus-
sock tundra, to shrub-poor tussock tundra. The shapes
of drift profiles in the lee of the shrubs were similar to