Spatial Heterogeneity in the Shrub
Tundra Ecotone in the Mackenzie
Delta Region, Northwest Territories:
Implications for Arctic
Environmental Change
Trevor C. Lantz,1* Sarah E. Gergel,2 and Steven V. Kokelj3
1School of Environmental Studies, University of Victoria, PO Box 3060 STN CSC, Victoria, British Columbia V8W 3R4, Canada;
2Department of Forest Science, Centre for Applied Conservation Research, University of British Columbia, 3041-2424 Main Mall,
Vancouver, British Columbia V6T 1Z4, Canada; 3Renewable Resources and Environment, Indian and Northern Affairs, Canada, Box
1500, 3rd Floor Bellanca Building, Yellowknife, Northwest Territories X1A-2R3, Canada
ABSTRACT
Growing evidence suggests that plant communities
in the Low Arctic are responding to recent increases
in air temperature. Changes to vegetation, partic-
ularly shifts in the abundance of upright shrubs,
can influence surface energy balance (albedo),
sensible and latent heat flux (evapotranspiration),
snow conditions, and the ground thermal regime.
Understanding fine-scale variability in vegetation
across the shrub tundra ecotone is therefore
essential as a monitoring baseline. In this article,
we use object-based classifications of airphotos to
examine changes in vegetation characteristics
(cover and patch size) across a latitudinal gradient
in the Mackenzie Delta uplands. This area is fre-
quently mapped as homogenous vegetation, but it
exhibits fine-scale variability in cover and patch
size. Our results show that the total area and size of
individual patches of shrub tundra decrease with
increasing latitude. The gradual nature of this
transition and its correlation with latitudinal vari-
ation in temperature suggests that the position of
the shrub ecotone will be sensitive to continued
warming. The impacts of vegetation structure on
ecological processes make improved understanding
of this heterogeneity critical to biophysical models
of Low Arctic ecosystems.
Key words: Low Arctic; object-oriented; object-
based; climate change; airphotos; vegetation clas-
sification.
INTRODUCTION
The Arctic is often described as one of the major
components in the Earth’s cooling system. It plays a
critical role in the global climate system by reflecting
incoming solar radiation and by radiating energy
gains transferred from the tropics (Chapin and others
2005; McGuire and others 2006). Northward tem-
perature decreases across this biome are accompanied
by changes in ecosystem properties including, com-
munity composition, vegetation structure, net pri-
mary productivity, heterotrophic respiration, carbon
storage, albedo, and permafrost conditions (Chapin
Received 15 April 2009; accepted 9 December 2009;
published online 22 January 2010
Author Contributions: TCL conceived study; TCL, SVK performed
research; TCL analyzed data; TCL, SEG, SVK wrote the article.
*Corresponding author; e-mail: tlantz@uvic.ca
Ecosystems (2010) 13: 194–204
DOI: 10.1007/s10021-009-9310-0

2010 Springer Science+Business Media, LLC
194

and others 2000; McGuire and others 2002; Thomp-
son and others 2004; Euskirchen and others 2007;
Burn and Kokelj 2009). Changes to Arctic vegetation
such as increases in shrub abundance that may result
from increasing temperature (Stafford and others
2000; Kaplan and others 2003; Hassol 2004; Jo-
hannessen and others 2004; Tape and others 2006;
Notaro and others 2007) therefore have the potential
to feedback to the global climate system (Chapin and
others 2005; McGuire and others 2006).
Despite a range of classification schemes, broad-
scale transitions in vegetation structure are similar
throughout the Circum-Arctic (Bliss and Matveyeva
1992; Walker 2000; Epstein and others 2004a). The
transition between the southernmost portion of the
Low Arctic and the northern Boreal Forest is fre-
quently referred to as the forest tundra (Payette and
others 2001; Sirois 1992). This ecotone consists of a
mosaic of forest and woodland interspersed with
shrub tundra and wetlands. Moving northward, trees
(typically Picea, Larix, Pinus, or Betula) give way to
tundra dominated by shrubs that are 40–400-cm tall.
This tall shrub zone is characterized by willows (Salix),
alder (Alnus), dwarf birches (Betula), and a mix of
ericaceous shrubs (Ledum, Vaccinium, and Arctostaph-
ylos). Elsewhere, this zone has been referred to as the
low shrub subzone (Walker and others 2002). At
higher latitudes terrain dominated by tall shrubs is
replaced by erect dwarf shrub tundra. Vegetation in
this physiognomic unit is less than 40-cm tall and is
characterized by dwarf shrubs (Betula, Salix, Vaccini-
um, Ledum, Empetrum, and Dryas) and sedges (Carex
and Eriophorum). Further north erect dwarf shrubs
are replaced by dwarf shrubs less than 10-cm tall.
Shrubs and forbs in this prostrate dwarf shrub zone
include: Cassiope, Dryas, Salix, Draba, Saxifraga, and
Carex. In the northernmost portion of the Arctic
biome landscapes typically have less than 5% vascu-
lar plant cover. Vegetation in this cushion-forb zone
consists of scattered bryophytes, cyanobacteria, small
forbs (Draba, Papaver, and Saxifraga), grasses (Puccin-
ellia and Alopecurus), and lichens. This study focuses
on the transition between the tall shrub zone and the
erect dwarf shrub zone. For simplicity we refer to
these zones throughout the manuscript as the shrub
tundra (tall shrub tundra) and dwarf shrub tundra
(erect dwarf shrub tundra).
At finer scales the transitions between Arctic
vegetation zones exhibit considerable heterogene-
ity. For example, in the tall shrub–dwarf shrub
tundra ecotone, changes in vegetation structure can
occur across scales as fine as 1 m (Bliss and Mat-
veyeva 1992; Walker and others 1994; Epstein and
others 2004b). Because the response of vegetation in
this ecotone to changes in climate will likely also
occur at fine scales, it will be difficult to detect
changes using land-cover classifications derived
from broad-scale satellite imagery. Despite the
importance of these fine-scale transitions, adequate
baseline data and monitoring strategies are lacking
in many areas (Bliss and Matveyeva 1992; Walker
and others 1994; Epstein and others 2004b).
The anticipated expansion of tall shrub tundra,
coupled with continued increases in air tempera-
ture are likely to have long-term impacts on per-
mafrost temperatures and terrain stability across the
Low Arctic (Sturm and others 2001a; Epstein and
others 2004b; Chapin and others 2005; McGuire
and others 2006). The accurate identification of the
transition from tall shrub to dwarf shrub tundra is
particularly critical to future predictions and track-
ing of northern environmental change because this
ecotone corresponds to large differences in albedo,
sensible heat flux, and duration and depth of snow
pack (Pomeroy and others 1995; Epstein and others
2004a; Chapin and others 2005; Sturm and others
2005). It has also been proposed that feedbacks
between vegetation, snow, ground heat flux, and
nutrient availability will accelerate the rate of veg-
etation change in the Low Arctic (Sturm and others
2005), and potentially lead to the warming of per-
mafrost. Because these feedbacks may be sensitive
to threshold patch sizes (Pomeroy and others 1995;
Sturm and others 2005) understanding variability
in patch size across this transition is also important.
In the uplands north of Inuvik, Northwest Ter-
ritories (NWT) there is a latitudinal shift from
tundra communities dominated by shrubs more
than 40-cm tall to those characterized by the
abundance of dwarf shrubs and sedges less than
40 cm (Mackay 1963; Corns 1974; Forest Man-
agement Institute 1975). Recent evidence suggests
that tall shrub tundra is encroaching into areas of
dwarf shrub tundra across the entire circumpolar
region (Silapaswan and others 2001; Sturm and
others 2001b; Stow and others 2004; Tape and
others 2006), but in the Mackenzie Delta Region,
base-line data on this transition are lacking. In this
article, we use object-based classification (Benz and
others 2004) of airphotos to describe fine-scale
changes in the proportion and patch sizes of shrub
tundra and dwarf shrub tundra across the shrub
tundra ecotone in the Mackenzie delta uplands.
METHODS
Study Area
Our study area in northwestern Canada is
approximately 11,000 km2 (Figure 1). This area
Spatial Heterogeneity in the Shrub Tundra Ecotone 195

east of the Mackenzie River delta is characterized
by subtle topography and thousands of small lakes
(Mackay 1963; Burn and Kokelj 2009). Quater-
nary surficial materials (primarily morainal
deposits) and soils (predominantly silty clays) are
relatively homogenous across the study area
(Mackay 1963; Aylsworth and others 2000; Soil
Landscapes of Canada Working Group 2007).
From October through April mean air tempera-
tures in the region are less than 0C. During the
short-growing season there is a linear temperature
gradient across the study area, where mean air
temperatures decrease from 9.4C (south) to 6.8C
(north) (Burn 1997; Lantz and others 2009;
Ritchie 1984). Air and ground temperatures in the
study region have increased in the last three
decades (Burn and Kokelj 2009) and have likely
contributed to observed increases in disturbances
associated with melting ground ice (Lantz and
Koklej 2008). The footprint of anthropogenic dis-
turbance is also anticipated to grow as exploration
and development intensify in the region (Holroyd
and Retzer 2005; Johnstone and Kokelj 2008;
Kemper and Macdonald 2009).
Figure 1. Map of the
study region showing the
study area, settlements,
water (light gray), and
airphoto study plots. Inset
map at the bottom right
shows the approximate
position of the study area
in North America.
196 T. C. Lantz and others

Airphoto Selection and Image
Manipulation
To describe the vegetation structure across the
uplands of the Mackenzie Delta region, we selected
18 air photos from a systematic survey of the
Mackenzie Delta region completed in 2004 (http://
www.gnwtgeomatics.nt.ca). We used a geodat-
abase of airphoto centers to randomly select 100
images between 6826¢ and 6934¢ (Figure 1).
From these photos we chose 18 that spanned the
study area, but had not been impacted by recent
fires or seismic exploration. We rejected photos of
areas with densities of seismic lines (areas where
seismic exploration vehicles have been driven)
exceeding 5 km/km2, or within 1 km of known
tundra fires. Each of the 18 images selected from
this survey covered an area of approximately
49 km2.
Negatives were scanned at 1814 dpi (1 pix-
el = 0.41 m) on a high-resolution photogrammet-
ric quality scanner. Scanned airphotos were
orthorectified using a 30-m DEM and Landsat 7
panchromatic orthoimagery (15-m pixel). Ortho-
rectification was performed using a nearest neigh-
bor algorithm (PCI Geomatics 2001) resulting in
root mean square errors generally less than 3 or-
thophoto pixels. Each orthorectified image covered
an area of approximately 49 km2, but was clipped
to a 36-km2 area surrounding the principal point
for classification and analysis. Images were also
tested for, and did not exhibit, systematic bias in
brightness (Lantz 2008).
Rationale for Object-Based Approach
To describe variability in vegetation structure and
patch size across the study area we used the
Definiens software package to perform object-based
classifications of each image (Definiens 2006). An
object-based approach to image classification differs
from conventional pixel-based methods by assign-
ing class membership to groups of pixels (objects)
rather than individual pixels. It is essentially a two-
step process that involves segmenting imagery into
image objects (groups of pixels) followed by the
classification of these objects. This approach en-
ables the multi-scale description of patch structure
and has been used successfully to map fine-scale
pattern on air photos of shrub-dominated ecosys-
tems (Laliberte and others 2004; Smith and others
2008). In the Definiens software package, seg-
mentation is a bottom-up region merging algo-
rithm that optimizes object creation by minimizing
the heterogeneity (color) of pixels contained in
each object, while creating objects that conform to
user-defined shape criteria (Definiens 2006). The
region merging process stops when the heteroge-
neity of an object exceeds a threshold defined by a
unitless scale parameter. User modification of this
threshold results in the creation of larger (higher
heterogeneity), or smaller objects (lower hetero-
geneity) (Benz and others 2004). By iteratively
modifying the heterogeneity threshold (by chang-
ing the scale parameter) users segment the image
into objects that reflect the structure of the land-
scape (Blaschke and Hay 2001; Definiens 2006). In
the Definiens software package, subsequent classi-
fication of image objects is accomplished using de-
fined membership rules (for example, thresholds)
or a nearest neighbor classification based on train-
ing data from the area of study (Laliberte and
others 2004).
Object-Based Segmentation,
Classification, and Accuracy Assessment
Segmentation
To minimize confusion between water- and dark-
colored shrub tundra in this lake-rich region, we
segmented each image at two scales. First, we
segmented images into large heterogeneous objects
whose borders corresponded to the boundary be-
tween water bodies and land (Figure 2). These
coarse-scale objects were created by performing
segmentation on the red, green, and blue bands, as
well as two texture measures. Textural co-occur-
rence measures (contrast and entropy) were cal-
culated using a grayscale band (ENVI 2006).
Segmentation was performed using a scale param-
eter of 600, a color to shape ratio of 0.9, and
compactness to smoothness ratio of 0.5 (Definiens
2006). Secondly, we performed a fine-scale seg-
mentation using a scale parameter that yielded
small homogenous objects that had shape and size
similar to isolated patches of shrub tundra or dwarf
shrub tundra (Figure 2). These fine-scale objects
were created by performing segmentation of the
red, green, and blue bands using a scale parameter
of 25, a color to shape to ratio of 0.9, and com-
pactness to smoothness of 0.5 (Definiens 2006).
Classification
After segmenting the images into objects, the objects
were classified into areas representing: (1) shrub
tundra [vegetation dominated by tall shrubs (Alnus
viridis, tall Salix spp. and Betula glandulosa)], (2) dwarf
shrub tundra [vegetation dominated by dwarf shrubs
(Ledum decumbens, Vaccinium vitis-idaea, Arctostaphylos
Spatial Heterogeneity in the Shrub Tundra Ecotone 197

rubra, Rubus chamaemorus) and sedges (Eriophorum
vaginatum, Kobresia hyperborea)], (3) water (lakes,
rivers, ponds, and the Beaufort Sea), and (4) bare
ground. We assigned fine-scale objects within the
broader ‘‘land’’ class to shrub tundra, dwarf shrub
tundra, smaller water bodies, or bare ground using a
nearest neighbor classifier. Fine-scale objects con-
tained within the broader ‘‘water’’ class were auto-
matically classified as water. To perform each
classification we used training data from high-reso-
lution ground truth images. Training images were
collected in the summer of 2006 using a Canon
PowerShot S80 digital camera mounted on a heli-
copter. Photographs were captured at an altitude of
approximately 450 m, had pixel sizes typically less
than 0.25 m (Figure 3), and were georeferenced in
ARCGIS using the 1:30,000 scale orthophotographs
of each plot. When classifications were complete, we
used Definiens to calculate the total areaof each cover
type and the proportion of shrub tundra in each
Figure 2. Diagram
showing the sequence of
operations in the object-
based classification of air
photo. Panel 1 aerial
photo of upland tundra
plot at coarse (1A) and
fine (1B) scales. Panel 2
segmentation of the
image at a coarse-scale
(2A) is followed by
segmentation at a fine-
scale (2B). Panel 3.
Coarse-scale objects are
classified into land and
water (3A) and fine-scale
objects are classified as
small water bodies, shrub
tundra, and dwarf shrub
tundra (4B), constrained
by their membership in
the coarse-scale
classification (4A).
198 T. C. Lantz and others

photo. We also used Definiens to merge all contigu-
ous objects of the same cover type and subsequently
to determine the total number of patches and mean
patch size for shrub tundra and dwarf shrub tundra
classes.
Accuracy Assessment
To examine the accuracy of our object-based clas-
sifications, we constructed confusion matrices
using two methods. In both the methods ground
truth data were derived from independent manual
classifications of five randomly selected images
collected in the same manner as our training
images. First, to conduct a standard pixel-based
accuracy assessment, we compared 1,500 random
points (300/ground truth image) from each cover
type in our classifications with classified ground
truth photos. To evaluate our estimates of patch
sizes, we also conducted a polygon-based accuracy
assessment (at the object level). To do this, we
compared 1,500 randomly selected objects from
each cover type with the ground-truth classifica-
tion. Bare ground occupied less than 0.05% of the
total area mapped and thus it was not feasible to
include this cover type in accuracy assessments. We
calculated overall accuracy, per class user’s, and
producer’s accuracies and the kappa statistic
(Lillesand and others 2003).
Statistical Analyses
To describe changes in the proportion and patch
sizes of shrub tundra and dwarf shrub tundra with
latitude we used regression analysis. We compared
linear and non-linear models of proportion and
patch size versus latitude by comparing Akaike’s
information criterion (AIC), AIC weights, and ad-
justed R2 values (Anderson and others 2000; R
Development Core Team 2006). We also examined
residual plots to ensure that models met the
assumptions of equal variance and normality. In
plots that had available temperature data (Lantz
and others 2009) we examined the Pearson corre-
lations coefficient between the proportion of shrub
tundra and mean summer (June–August) temper-
ature.
RESULTS
Latitudinal Changes in the Proportion
and Patch Sizes of Shrub Tundra and
Dwarf Shrub Tundra
There was a northward decrease in the propor-
tional area and mean patch size of shrub tundra
across the study area (Figure 4). The northward
decrease in the dominance of shrub tundra corre-
sponded to an increase in the abundance and patch
size of dwarf shrub tundra (Figure 5). All models
showed strong evidence of non-linear relationships
and non-linear models had improved fit and lower
AIC’s (Table 1). In all models the relationships be-
tween latitude and patch size and proportion of
cover showed a steeper relationship north of
68.9N. The Pearson correlation coefficient be-
tween the proportion of shrub tundra and mean
summer temperatures was 0.95 and the latitude at
which the proportion of shrub tundra declined
below 50% corresponded approximately to the
mean 10C July isotherm in the region (Pelletier
n.d.).
Accuracy Assessment
Pixel- and object-based estimates of overall classi-
fication accuracy were 85.8 and 78.1%, respec-
tively. The kappa statistic, which ranges from 0 to 1
and provides an estimate of overall accuracy that
accounts for the possibility that objects will be
correctly classified by chance (Lillesand and others
2003), was 0.787 for the pixel-based method and
0.664 for the polygon-based method (Table 2).
User accuracies (the percent of map units tested
that were the same as the truth data) indicate that
water bodies were extremely well classified (user
accuracies 93.3–96.7%). Producer accuracies (the
percent of truth points that were mapped correctly)
were also high for the water class (94.0–97.6%).
Conversely, shrub tundra and dwarf shrub tundra
classes were prone to some classification error.
Figure 3. Example image from helicopter surveys used to
train image classifiers and to conduct accuracy assess-
ments. Image shows shrub tundra (dark gray [green and
yellow]), dwarf shrub tundra (light gray [lavender-gray]),
and two small lakes (black [blue]). (Color figure online)
Spatial Heterogeneity in the Shrub Tundra Ecotone 199

(Tape and others 2006). Plot level manipulations of
temperature and nutrient availability further sup-
port predictions that warming will increase shrub
dominance in the Low Arctic (Parsons and others
1994; Chapin and others 1995; Bret-Harte and
others 2001; Bret-Harte and others 2002; Dormann
and Woodin 2002; Walker and others 2006).
Monitoring Shrub Encroachment
Accurate maps representing the shrub tundra
transition are critical to monitoring the rate of shrub
expansion in the Low Arctic. Our results are con-
sistent with previous descriptions of vegetation
transitions in the study area (Corns 1974; Forest
Management Institute 1975; IEG 2002), but are a
significant improvement to the accuracy of fine-
scale mapping of regional tundra vegetation struc-
ture. Here, we mapped vegetation structure of the
shrub–tundra transition in the Mackenzie Delta
uplands with an overall accuracy of 86% and user’s
accuracies ranging from 69 to 84% for shrub classes.
Previous classifications of the region had per class
accuracies as low as 50% for shrub dominated ter-
rain (IEG 2002). Estimates of the areal expansion of
tall shrubs in the Western Arctic vary between
approximately 1 and 6% per decade (Tape and
others 2006; Lantz, unpublished data). Conse-
quently, surveys repeated every 20–30 years using
the methods implemented here will be capable of
Table 1. Comparison of Linear and Non-Linear Models Using Adjusted R2, AIC, and AIC Weights
Dependant variable Model AIC DAIC Adjusted R2 AIC weight
Proportion shrub tundra Latitude -85.82 4.97 0.740 0.071
Latitude2 -85.89 4.90 0.741 0.074
Latitude + Latitude2 -90.78 0 0.811 0.855
Proportion dwarf shrub tundra Latitude -85.48 5.12 0.740 0.067
Latitude2 -85.55 5.04 0.741 0.069
Latitude + Latitude2 -90.60 0 0.813 0.864
Shrub tundra patch size Latitude 165.90 2.08 0.690 0.305
Latitude2 165.85 2.03 0.691 0.313
Latitude + Latitude2 163.82 0 0.737 0.864
Dwarf shrub tundra patch size Latitude 249.79 3.87 0.530 0.125
Latitude2 249.74 3.83 0.532 0.127
Latitude + Latitude2 245.92 0 0.639 0.864
Best models (shown in bold) are plotted in Figures 4 and 5.
Table 2. Classification Accuracy Assessments
Classified data Ground truth data
Water Shrub tundra Dwarf shrub tundra User’s accuracies (%)
Pixel-based
Water 1451 31 18 96.7
Shrub tundra 31 1154 315 76.9
Dwarf shrub tundra 4 239 1257 83.8
Producers’ accuracies (%) 97.6 81.0 79.1
Overall accuracy (%) 85.8
Kappa coefficient 0.787
Object-based
Water 850 27 34 93.3
Shrub tundra 44 1163 337 75.3
Dwarf shrub tundra 10 319 740 69.2
Producers’ accuracies (%) 94.0 77.1 66.6
Overall accuracy (%) 78.1
Kappa coefficient 0.664
Table shows raw tallies, producers and user’s accuracies, overall accuracy, and the kappa coefficient.
Spatial Heterogeneity in the Shrub Tundra Ecotone 201

detecting rapid shrub expansion, but may not be
suitable for tracking change in areas where tall
shrub expansion is slow. Efforts to track broad-scale
changes in tall shrubs using remote sensing should
therefore be coupled with ground-based surveys.
Improvements in classification accuracy would
make it possible to use this technique to detect finer-
scale spatial and temporal changes in vegetation
structure. For example, object-based classifications
combining fine-scale airphotos and high-resolution
multispectral imagery would likely improve accu-
racy. Additional contextual layers, including eleva-
tion and object texture, could also be used to further
refine object-based classifications of these ecosys-
tems (Dorren and others 2003; Bock and others
2005).
Implications
To date, there has been insufficient research
describing and mapping the transition between
shrub tundra and dwarf shrub tundra in the Mac-
kenzie Delta uplands. Air and ground temperatures
in this region are warming and the frequency of
natural and anthropogenic disturbance is increas-
ing (Lantz and Kokelj 2008; Burn and Kokelj 2009;
Johnstone and Kokelj 2008; Kemper and Mac-
donald 2009). Disentangling and tracking the ef-
fects of multiple perturbations on the vegetation
across this ecotone requires an accurate baseline.
Although frequently mapped as a homogeneous
cover type (Gould and others 2002; Walker and
others 2002; Gould and others 2003) vegetation in
this region shows significant non-linear changes in
the relative abundance and patch size of shrub
tundra and dwarf shrub tundra with increasing
latitude. Strong correlations between this vegeta-
tion transition and regional temperature, coupled
with evidence of increases in tall shrub tundra in
other regions, suggest that further warming is likely
to alter the structure of this ecotone.
The relations between vegetation structure,
snow cover, and permafrost conditions (Burn and
Kokelj 2009), highlight the importance of under-
standing vegetation change in this region. Large
differences in the properties of tall shrub and dwarf
shrub tundra, including snowpack, the duration of
the snow-free season, albedo, methane flux, and
active layer thickness (Pomeroy and others 1997;
Epstein and others 2004a; Chapin and others 2005;
Sturm and others 2005), suggest that changes in
shrub abundance across this ecotone may alter
ecosystem function (Epstein and others 2004a;
Chapin and others 2005; Sturm and others 2005)
and impact the ground thermal regime and terrain
stability (Burn and Kokelj 2009; Kokelj and others
2009; Lantz and others 2009). Accurate maps of
fine-scale differences in vegetation structure are
therefore essential for establishing a baseline from
which to track change, and for realistically
parameterizing regional models of ecosystem pro-
cesses (Nelson and others 1997; Reeburgh and
others 1998; Oechel and others 2000; Chapin and
others 2002; Schneider and others 2009).
ACKNOWLEDGMENTS
The authors thank Sarah Borgart, Stephen Schwarz,
Marcella Snijders, Matt Tomlinson, and Rory
Tooke. We would also like to thank Greg Henry, Isla
Myers-Smith, Nicholas Coops, and two anonymous
reviewers for helpful comments on drafts of this
manuscript. Funding support was received from
Aurora Research Institute (Research Fellowship),
Canon USA and the AAAS (Canon National Parks
Science Scholarship), Global Forest Research (Re-
search Grant GF-18-2004-212), Indian and North-
ern Affairs Canada (Cumulative Impact Monitoring
Program, Water Resources Division, the Northern
Science Training Program, and the Mackenzie Val-
ley Airphoto Project), Killiam Trusts (Predoctoral
Fellowship), Natural Resources Canada (Polar
Continental Shelf Program), Natural Sciences and
Engineering Research Council of Canada (PGS-B
and Northern Internship to T. C Lantz).
REFERENCES
Anderson DR, Burnham KP, Thompson WL. 2000. Null
hypothesis testing: problems, prevalence, and an alternative.
J Wildlife Manag 64:912–23.
Aylsworth JM, Burgess MM, Desrochers DT, Duk-Rodkin A,
Robertson T, Traynor JA. 2000. Surficial geology, subsurface
materials, and thaw sensitivity of sediments. In: Dyke LD,
Brooks GR, Eds. The physical environment of the Mackenzie
Valley, Northwest Territories: a base line for the assessment of
environmental change. Geological Survey of Canada Bulletin,
vol 547. Ottawa, ON: Geological Survey of Canada. p 41–8.
Benz UC, Hofmann P, Willhauck G, Lingenfelder I, Heynen M.
2004. Multi-resolution, object-oriented fuzzy analysis of re-
mote sensing data for GIS-ready information. ISPRS J Pho-
togramm Remote Sens 58:239–58.
Blaschke T, Hay GJ. 2001. Object-oriented image analysis and
scale-space: theory and methods for modeling and evaluating
multiscale landscape structure. Int Arch Photogramm Remote
Sens 34:22–9.
Bliss LC, Matveyeva NV. 1992. Circumpolar Arctic vegetation.
In: Chapin FS, Ed. Arctic ecosystems in a changing climate: an
ecophysiological perspective. San Diego: Academic Press.
p 59–89.
Bock M, Xofis P, Mitchley J, Rossner G, Wissen M. 2005. Object-
oriented methods for habitat mapping at multiple scales—case
studies from Northern Germany and Wye Downs, UK. J Nat
Conserv 13:75–89.
202 T. C. Lantz and others

Bret-Harte MS, Shaver GR, Chapin FS. 2002. Primary and sec-
ondary stem growth in arctic shrubs: implications for com-
munity response to environmental change. J Ecol 90:251–67.
Bret-Harte MS, Shaver GR, Zoerner JP, Johnstone JF, Wagner JL,
Chavez AS, Gunkelman RF, Lippert SC, Laundre JA. 2001.
Developmental plasticity allows Betula nana to dominate tun-
dra subjected to an altered environment. Ecology 82:18–32.
Burn CR. 1997. Cryostratigraphy, paleogeography, and climate
change during the early Holocene warm interval, western
Arctic coast, Canada. Can J Earth Sci 34:912–25.
Burn CR, Kokelj SV. 2009. The environment and permafrost of
the Mackenzie Delta area. Permafrost Periglac Process 20:83–
105. doi:10.1002/ppp.642.
Chapin FS, Eugster W, McFadden JP, Lynch AH, Walker DA.
2002. Summer differences among Arctic ecosystems in
regional climate forcing. J Clim 13:2002–10.
Chapin FS, McGuire AD, Randerson J, Pielke R, Baldocchi D,
Hobbie SE, Roulet N, Eugster W, Kasischke E, Rastetter EB,
Zimov SA, Running SW. 2000. Arctic and boreal ecosystems
of western North America as components of the climate sys-
tem. Global Change Biol 6:211–23.
Chapin FS, Shaver GR, Giblin AE, Nadelhoffer KJ, Laundre JA.
1995. Responses of Arctic tundra to experimental and
observed changes in climate. Ecology 76:694–711.
Chapin FS, Sturm M, Serreze MC, McFadden JP, Key JR, Lloyd
AH, McGuire AD, Rupp TS, Lynch AH, Schimel JP, Beringer J,
Chapman WL, Epstein HE, Euskirchen ES, Hinzman LD, Jia G,
Ping CL, Tape KD, Thompson CDC, Walker DA, Welker JM.
2005. Role of land-surface changes in Arctic summer warm-
ing. Science 310:657–60.
Corns IGW. 1974. Arctic plant communities east of Mackenzie-
Delta. Can J Bot 52:1731–45.
Definiens. 2006. Definiens Professional 5 User Guide. Munich:
Definiens AG.
Definiens. 2007. Definiens Developer 7 User Guide. Munich:
Definiens AG.
Dormann CF, Woodin SJ. 2002. Climate change in the Arctic:
using plant functional types in a meta-analysis of field
experiments. Funct Ecol 16:4–17.
Dorren LKA, Maier B, Seijmonsbergen AC. 2003. Improved
Landsat-based forest mapping in steep mountainous terrain
using object-based classification. For Ecol Manag 183:31–46.
ENVI. 2006. ENVI User’s Guide. Version 4.3. Boulder, CO:
Research Systems, Inc.
Epstein HE, Beringer J, Gould WA, Lloyd AH, Thompson CD,
Chapin FS, Michaelson GJ, Ping CL, Rupp TS, Walker DA.
2004a. The nature of spatial transitions in the Arctic. J Bio-
geogr 31:1917–33.
Epstein HE, Calef MP, Walker MD, Chapin FS, Starfield AM.
2004b. Detecting changes in Arctic tundra plant communities
in response to warming over decadal time scales. Global
Change Biol 10:1325–34.
Euskirchen ES, Mcguire AD, Chapin FS. 2007. Energy feedbacks
of northern high-latitude ecosystems to the climate system
due to reduced snow cover during 20th century warming.
Global Change Biol 13:2425–38.
Forest Management Institute. 1975. Vegetation types of the
Lower Mackenzie and Yukon corridor. Report No. 74-40.
Canadian Forest Service, Ottawa, ON.
Gould WA, Edlund S, Zoltai S, Raynolds M, Walker DA, Maier H.
2002. Canadian Arctic vegetation mapping. Int J Remote Sens
23:4597–609.
Gould WA, Raynolds M, Walker DA. 2003. Vegetation, plant
biomass, and net primary productivity patterns in the Cana-
dian Arctic. J Geophys Res Atmos 108(D2):8167. doi:10.1029/
2001JD000948.
Hassol SJ, Ed. 2004. Arctic climate impact assessment: impacts of
warming climate. Cambridge: Cambridge University Press.
Holroyd P, Retzer H. 2005. A peak into the future: the potential
landscape impacts from gas development in Northern Canada.
Calgary,AB: ThePembina Institute for AppropriateDevelopment.
IEG. 2002. Vegetation classification and wildlife habitat suit-
ability modeling in the Mackenzie Delta Region. Project
#5003–01. Calgary, AB: Inuvialuit Environmental and
Geotechnical Inc.
Johannessen OM, Bengtsson L, Miles MW, Kuzmina SI, Seme-
nov VA, Alekseev GV, Nagurnyi AP, Zakharov VF, Bobylev
LP, Pettersson LH, Hasselmann K, Cattle AP. 2004. Arctic
climate change: observed and modelled temperature and
sea-ice variability. Tellus Ser A 56:328–41.
Johnstone JF, Kokelj SV. 2008. Environmental conditions and
vegetation recovery at abandoned-drilling mud sumps in the
Mackenzie Delta region, NWT, Canada. Arctic 61:199–211.
Kaplan JO, Bigelow NH, Prentice IC, Harrison SP, Bartlein PJ,
Christensen TR, Cramer W, Matveyeva NV, McGuire AD,
Murray DF, Razzhivin VY, Smith B, Walker DA, Anderson
PM, Andreev AA, Brubaker LB, Edwards ME, Lozhkin AV.
2003. Climate change and Arctic ecosystems: 2. Modeling,
paleodata-model comparisons, and future projections. J Geo-
phys Res Atmos 108(D19):8171. doi:10.1029/2002JD002559.
Laliberte AS, Rango A, Havstad KM, Paris JF, Beck RF, McNeely
R, Gonzalez AL. 2004. Object-oriented image analysis for
mapping shrub encroachment from 1937 to 2003 in southern
New Mexico. Remote Sens Environ 93:198–210.
Kemper JT, Macdonald SE. 2009. Effects of contemporary
Winter seismic exploration on Low Arctic plant communities
and permafrost. Arct Antarct Alp Res 41:228–37.
Kokelj SV, Lantz TC, Kanigan J, Smith SL, Coutts R. 2009. Origin
and polycyclic behavior of tundra thaw slumps, Mackenzie
Delta region, Northwest Territories, Canada. Permafrost
Periglac Process 20:173–84. doi:10.1002/ppp.642.
Lantz TC. 2008. Relative influence of temperature and distur-
bance on vegetation dynamics in the Low Arctic: an investi-
gation at multiple scales. PhD Dissertation, University of
British Columbia, Vancouver.
Lantz TC, Kokelj SV. 2008. Increasing rates of retrogressive thaw
slump activity in the Mackenzie Delta region, N.W.T., Canada.
Geophys Res Lett 35:L06502. doi:06510.01029/02007GL032
433.
Lantz TC, Kokelj SVK, Gergel SE, Henry GHR. 2009. Relative
impacts of disturbance and temperature: persistent changes in
microenvironment and vegetation in retrogressive thaw
slumps. Global Change Biol 15:1664–75. doi:10.1111/j.1365-
2486.2009.01917.x.
Lillesand TM, Kiefer RW, Chipman JW. 2003. Remote sensing
and image interpretation. 5th edn. New York: Wiley.
Mackay JR. 1963. The Mackenzie Delta Area, N.W.T., Geo-
graphical Branch Memoir 8. Ottawa, ON: Department of
Mines and Technical Surveys.
McGuire AD, Chapin FS, Walsh JE, Wirth C. 2006. Integrated
regional changes in Arctic climate feedbacks: implications for
the global climate system. Ann Rev Environ Resour 31:61–91.
McGuire AD, Wirth C, Apps M, Beringer J, Clein J, Epstein H,
Kicklighter DW, Bhatti J, Chapin FS, de Groot B, Efremov D,
Spatial Heterogeneity in the Shrub Tundra Ecotone 203

Eugster W, Fukuda M, Gower T, Hinzman L, Huntley B, Jia
GJ, Kasischke E, Melillo J, Romanovsky V, Shvidenko A,
Vaganov E, Walker D. 2002. Environmental variation, vege-
tation distribution, carbon dynamics and water/energy ex-
change at high latitudes. J Veg Sci 13:301–14.
Muller SV, Racoviteanu AE, Walker DA. 1999. Landsat MSS-
derived land-cover map of northern Alaska: extrapolation
methods and a comparison with photo-interpreted and AV-
HRR-derived maps. Int J Remote Sens 20:2921–46.
Nelson FE, Shiklomanov NI, Mueller GR, Hinkel KM, Walker
DA, Bockheim JG. 1997. Estimating active-layer thickness
over a large region: Kuparuk River Basin, Alaska, USA. Arctic
Alpine Res 29:367–78.
Notaro M, Vavrus S, Liu ZY. 2007. Global vegetation and climate
change due to future increases in CO2 as projected by a fully
coupled model with dynamic vegetation. J Clim 20:70–90.
Oechel WC, Vourlitis GL, Verfaillie J, Crawford T, Brooks S, Du-
mas E, Hope A, Stow D, Boynton B, Nosov V, Zulueta R. 2000.
A scaling approach for quantifying the net CO2 flux of the
Kuparuk River Basin, Alaska. Global Change Biol 6:160–73.
Olthof I, Pouliot D, Latifovic R, Chen WJ. 2008. Recent (1986–
2006) Vegetation-specific NDVI trends in Northern Canada
from satellite data. Arctic 61:381–94.
Parsons AN, Welker JM, Wookey PA, Press MC, Callaghan TV,
Lee JA. 1994. Growth responses of 4 Sub-Arctic dwarf shrubs
to simulated environmental change. J Ecol 82:307–18.
Payette S, Fortin MJ, Gamache I. 2001. The subarctic forest-
tundra: the structure of a biome in a changing climate. Bio-
science 51:709–18.
PCI Geomatics. 2001. OrthoEngine Reference Manual. Version
8.2. Richmond Hill, ON: PCI Geomatics.
Pelletier BR. n.d. Environmental Atlas of the Beaufort Coast-
lands. Geological Survey of Canada. http://gsc.nrcan.gc.ca/
beaufort/index_e.php. Accessed December 2009.
Pomeroy JW, Marsh P, Gray DM. 1997. Application of a dis-
tributed blowing snow model to the Arctic. Hydrol Process
11:1451–64.
Pomeroy JW, Marsh P, Jones HG, Davies TD. 1995. Spatial dis-
tribution of snow chemical load at the tundra-taiga transition.
In: Tonnessen KA, Williams MW, Tranter M, Eds. Biogeo-
chemistry of seasonally snow-covered catchments. Interna-
tional Association of Hydrological Sciences No. 228.
Wallingford: IAHS Press. p 191–203.
R Development Core Team. 2006. R: a language and environ-
ment for statistical computing, reference index, version 2.6.2.
R Foundation for Statistical Computing. http://www.R-pro-
ject.org, Vienna, Austria.
Reeburgh WS, King JY, Regli SK, Kling GW, Auerbach NA,
Walker DA. 1998. A CH4 emission estimate for the Kuparuk
River basin, Alaska. J Geophys Res Atmos 103:29005–13.
Richards JA, Jia X. 2006. Remote sensing digital image analysis:
an introduction. Heidelberg: Springer.
Ritchie JC. 1984. Past and present vegetation of the far North-
west of Canada. Toronto, ON: University of Toronto Press.
Schimel JP, Bilbrough C, Welker JA. 2004. Increased snow
depth affects microbial activity and nitrogen mineralization in
two Arctic tundra communities. Soil Biol Biochem 36:217–27.
Schneider J, Grosse G, Wagner D. 2009. Land cover classification
of tundra environments in the Arctic Lena Delta based on
Landsat 7 ETM+ data and its application for upscaling of
methane emissions. Remote Sens Environ 113:380–91.
Silapaswan CS, Verbyla DL, McGuire AD. 2001. Land cover
change on the Seward Peninsula: the use of remote sensing to
evaluate the potential influences of climate warming on his-
torical vegetation dynamics. Can J Remote Sens 27:542–54.
Sirois L. 1992. The transition between boreal forest and tundra.
In: Shugart HH, Leemans R, Bonan GB, Eds. A systems
analysis of the global boreal forest. Cambridge University
Press: Cambridge, p 197–215
Smith A, Strand E, Steele C, Hann D, Garrity S, Falkowski M,
Evans J. 2008. Production of vegetation spatial-structure maps
by per-object analysis of juniper encroachment in multitem-
poral aerial photographs. Can J Remote Sens 34:S268–85.
Soil Landscapes of Canada Working Group. 2007. Soil Landscapes
of Canada v3.1.1 (digital map and database at 1:1 million scale).
Ottawa, ON: Agriculture and Agri-Food Canada.
Stafford JM, Wendler G, Curtis J. 2000. Temperature and pre-
cipitation of Alaska: 50 year trend analysis. Theor Appl Cli-
matol 67:33–44.
Stow DA, Hope A, McGuire D, Verbyla D, Gamon J, Huemmrich
F, Houston S, Racine C, Sturm M, Tape K, Hinzman L, Yos-
hikawa K, Tweedie C, Noyle B, Silapaswan C, Douglas D,
Griffith B, Jia G, Epstein H, Walker D, Daeschner S, Petersen
A, Zhou LM, Myneni R. 2004. Remote sensing of vegetation
and land-cover change in Arctic tundra ecosystems. Remote
Sens Environ 89:281–308.
Sturm M, McFadden JP, Liston GE, Chapin FS, Racine CH,
Holmgren J. 2001a. Snow-shrub interactions in Arctic tundra:
a hypothesis with climatic implications. J Clim 14:336–44.
Sturm M, Racine C, Tape K. 2001b. Climate change—increasing
shrub abundance in the Arctic. Nature 411:546–7.
Sturm M, Schimel J, Michaelson G, Welker JM, Oberbauer SF,
Liston GE, Fahnestock J, Romanovsky VE. 2005. Winter bio-
logical processes could help convert Arctic tundra to shrub
land. Bioscience 55:17–26.
Tape K, Sturm M, Racine C. 2006. The evidence for shrub
expansion in Northern Alaska and the Pan-Arctic. Global
Change Biol 12:686–702.
Thompson C, Beringer J, Chapin FS, McGuire AD. 2004.
Structural complexity and land-surface energy exchange
along a gradient from Arctic tundra to boreal forest. J Veg Sci
15:397–406.
Walker DA. 2000. Hierarchical subdivision of Arctic tundra
based on vegetation response to climate, parent material and
topography. Global Change Biol 6:19–34.
Walker DA, Gould WA, Maier HA, Raynolds MK. 2002. The
circumpolar Arctic vegetation map: AVHRR-derived base
maps, environmental controls, and integrated mapping pro-
cedures. Int J Remote Sens 23:4551–70.
Walker MD, Wahren CH, Hollister RD, Henry GHR, Ahlquist LE,
Alatalo JM, Bret-Harte MS, Calef MP, Callaghan TV, Carroll
AB, Epstein HE, Jonsdottir IS, Klein JA, Magnusson B, Molau
U, Oberbauer SF, Rewa SP, Robinson CH, Shaver GR, Suding
KN, Thompson CC, Tolvanen A, Totland O, Turner PL,
Tweedie CE, Webber PJ, Wookey PA. 2006. Plant community
responses to experimental warming across the tundra biome.
Proc Natl Acad Sci USA 103:1342–6.
Walker MD, Walker DA, Auerbach NA. 1994. Plant communities
of a tussock tundra landscape in the Brooks range foothills,
Alaska. J Veg Sci 5:843–66.
204 T. C. Lantz and others