Content uploaded by F Stuart Chapin III
Author content
All content in this area was uploaded by F Stuart Chapin III on Nov 23, 2015
Content may be subject to copyright.
Content uploaded by Rebecca E. Hewitt
Author content
All content in this area was uploaded by Rebecca E. Hewitt on Nov 16, 2015
Content may be subject to copyright.
RESEARCH ARTICLE
Getting to the root of the matter: landscape implications
of plant-fungal interactions for tree migration in Alaska
Rebecca E. Hewitt .Alec P. Bennett .Amy L. Breen .
Teresa N. Hollingsworth .D. Lee Taylor .
F. Stuart Chapin III .T. Scott Rupp
Received: 8 May 2015 / Accepted: 29 October 2015
!Springer Science+Business Media Dordrecht 2015
Abstract
Context Forecasting the expansion of forest into
Alaska tundra is critical to predicting regional ecosys-
tem services, including climate feedbacks such as
carbon storage. Controls over seedling establishment
govern forest development and migration potential.
Ectomycorrhizal fungi (EMF), obligate symbionts of
all Alaskan tree species, are particularly important to
seedling establishment, yet their significance to land-
scape vegetation change is largely unknown.
Objective We used ALFRESCO, a landscape model
of wildfire and vegetation dynamics, to explore
whether EMF inoculum potential influences patterns
of tundra afforestation and associated flammability.
Methods Using two downscaled CMIP3 general
circulation models (ECHAM5 and CCCMA) and a
mid-range emissions scenario (A1B) at a 1 km
2
resolution, we compared simulated tundra afforesta-
tion rates and flammability from four parameteriza-
tions of EMF effects on seedling establishment and
growth from 2000 to 2100.
Results Modeling predicted an 8.8–18.2 % increase
in forest cover from 2000 to 2100. Simulations that
explicitly represented landscape variability in EMF
inoculum potential showed a reduced percent change
afforestation of up to a 2.8 % due to low inoculum
potential limiting seedling growth. This reduction
limited fuel availability and thus, cumulative area
burned. Regardless of inclusion of EMF effects in
simulations, landscape flammability was lower for
simulations driven by the wetter and cooler CCCMA
model than the warmer and drier ECHAM5 model,
while tundra afforestation was greater.
Conclusions Results suggest abiotic factors are the
primary driver of tree migration. Simulations includ-
ing EMF effects, a biotic factor, yielded more
conservative estimates of land cover change across
Alaska that better-matched empirical estimates from
the previous century.
Keywords Alaska !ALFRESCO !Climate change !
Ectomycorrhizal fungi !Treeline !Wildfire
R. E. Hewitt (&)!A. P. Bennett !A. L. Breen !
T. S. Rupp
International Arctic Research Center, Scenarios Network
for Alaska & Arctic Planning, University of Alaska
Fairbanks, P.O. Box 757245, Fairbanks, AK 99775, USA
e-mail: rebecca.hewitt@nau.edu
T. N. Hollingsworth
US Forest Service PNW Research Station, University of
Alaska Fairbanks, P.O. Box 756780, Fairbanks,
AK 99775, USA
D. L. Taylor
Department of Biology, University of New Mexico, 167
Castetter Hall, Albuquerque, NM 87131, USA
F. S. Chapin III
Institute of Arctic Biology, University of Alaska
Fairbanks, P.O. Box 756000, Fairbanks, AK 99775, USA
123
Landscape Ecol
DOI 10.1007/s10980-015-0306-1
Introduction
Soil microbes are critical to plant establishment,
survival, and growth (Horton and van der Heijden
2008; van der Heijden and Horton 2009; Bever et al.
2010), but rarely has the link between plant-fungal
interactions and landscape vegetation change been
explored. This is likely due to issues of scale, where
microbial composition can vary on the micro-scale
(Taylor et al. 2010) and vegetation composition can
vary on the landscape or regional scale (Turner 1989).
For example, controls over seedling establishment at
and beyond current treeline govern both stand devel-
opment at the range limit of the boreal forest and the
potential for migration (Hobbie and Chapin 1998;
Harsch and Bader 2011). However, the importance of
biotic factors, such as the effect of mycobionts on tree
seedling establishment, in understanding treeline
dynamics has been largely overlooked. Regional
changes in the cover of boreal forest and tundra
vegetation can influence the climate system through
changes in albedo and carbon storage (McGuire et al.
2001; Chapin et al. 2005; Euskirchen et al. 2009a).
Therefore, understanding the ecological factors that
influence the position of arctic treeline, the boundary
between the boreal forest and tundra, has strong
implications for accurately forecasting changes in
these ecosystem services.
Ectomycorrhizal fungi (EMF) are obligate sym-
bionts of all boreal tree species in Alaska. These
mycobionts are physiologically important to plant
performance because they are the conduits of soil
resources for host plants, especially during the
vulnerable seedling establishment stage (Horton and
van der Heijden 2008; Smith and Read 2008). Fungal
inoculum is generally not considered limited in soils
(but see Peay et al. 2010a). However, at Alaskan
treeline low availability of boreal tree compatible
fungal inoculum could limit seedling establishment
(Hewitt 2014). The transition from boreal forest to
tundra across the ecotone corresponds with a shift
from dominance of EMF host plants, such as aspen
(Populus tremuloides), spruce (Picea sp.), and birch
(Betula neoalaskana) trees, to ERM (ericoid mycor-
rhizal) host plants, such as blueberry and cranberry
(Vaccinium species) dwarf shrubs, and non-mycor-
rhizal or AM (arbuscular mycorrhizal) graminoid host
plants, with varying densities of EMF dwarf shrubs,
such as willow and birch (Read 1991; Gardes and
Dahlberg 1996). EMF composition and abundance has
been related to successful seedling establishment both
within and beyond the current range limit of host
plants in other ecosystems (Perry et al. 1982,1989;
Horton et al. 1999; Nara 2006; Nunez et al. 2009).
Across biomes, EMF richness and colonization of
seedlings declines with increased distance from forest
edge for both native and invasive tree species across
fine spatial scales (\1000 m) (Dickie and Reich 2005;
Nunez et al. 2009; Peay et al. 2010b,2012). These
isolation-effects on fungal communities persist
through stand development (Peay et al. 2010b), which
in turn could influence growth of early life-stages of
boreal trees that establish beyond current treeline.
However, as far as we know, a scaling exercise linking
biogeographic patterns of mycobionts to root tips,
seedling performance, and forecasts of landscape
vegetation transitions has not been accomplished until
this study.
Projected changes in treeline position are primarily
based on the assumption that high latitudinal and
altitudinal forests will respond positively to increases
in growing season and air temperatures (Harsch et al.
2009). Changes in fire regime in the boreal forest and
tundra biomes are also directly related to a warming
climate (Hu et al. 2010; Kelly et al. 2013) and may be a
greater driver of species migrations than temperature
per se (Dale et al. 2001). In Alaska and western Yukon
Territory, fire regime is tightly coupled with tree
seedling recruitment, forest species composition, and
the northward migration of lodgepole pine (Johnstone
and Chapin 2003,2006; Johnstone et al. 2010). This
has led to the hypothesis that fire will likely facilitate
afforestation of tundra by killing extant plant com-
petitors and opening up novel, high-quality microsites
for establishment (Landhausser and Wein 1993). In
contrast, mycobiont communities are often altered
post-fire when host plants are killed or there is severe
combustion of the upper soil horizons where they are
abundant (Dahlberg 2002; Cairney and Bastias 2007;
Hewitt et al. 2013; Hewitt 2014). These effects of fire
severity on fungal composition can in turn reduce
seedling performance (Hewitt 2014).
To explore the impacts of EMF on tree migration in
the tundra zone of Alaska we used a landscape model
of vegetation dynamics and fire activity. We per-
formed model simulation experiments using the
landscape model ALaska FRame-based EcoSystem
COde (ALFRESCO) (Rupp et al. 2000,2001) to
Landscape Ecol
123
quantify the effects of variability in EMF inoculum
potential on state transitions from tundra to forest with
climate warming and fire activity. Model simulation
results contribute to understanding how mycorrhizal
interactions and variability in landscape inoculum
potential may influence vegetation transitions and the
scope of potential changes in landscape vegetation
patterns, landscape flammability, and associated
ecosystem services with climate warming at high
latitudes.
Methods
Model overview
ALFRESCO is a frame-based spatially explicit state–
and-transition model (Starfield et al. 1993) that
simulates vegetation succession and fire occurrence
across Alaska at a 1 km
2
resolution and annual time
step. The model was originally developed to simulate
vegetation state transitions with changes in climate
(temperature and precipitation) and fire disturbance
(Rupp et al. 2000). The model assumes that vegetation
responds primarily to transient changes in climate and
fire regime (Starfield and Chapin 1996; Fig. 1).
Previous renditions of the model quantified spruce
migration in response to changes in climate and fire
(Breen et al. 2013; Rupp et al. 2001). Here we
developed a mycorrhizal-effects submodel that mod-
ifies seedling establishment and growth in relation to
transient changes in climate and fire activity to assess
the landscape implications of plant-fungal interactions
on tree migration.
ALFRESCO simulates wildfire as a stochastic
process. It uses a cellular automaton approach for
ignition of a 1 km
2
cell in the model and spread to the
surrounding cells. Ignition and spread are determined
by the comparison of a randomly generated number
and the flammability coefficient of a cell. Flammabil-
ity is determined by vegetation classification of a cell,
age of a cell, and climate parameters. Climate-effects
on flammability are a function of weighted monthly
mean temperatures and precipitation for the fire season
derived from a regression model, for the months of
March-September (Rupp et al. 2015).
ALFRESCO simulates vegetation succession and
state transitions as deterministic processes. State
transitions between vegetation classes are based on
rules for each vegetation type. The initial land cover
dataset used for ALFRESCO simulations is a modified
version from the North American Land Change
Monitoring System from 2005, with aggregated veg-
etation classes. For earlier versions of ALFRESCO,
successional dynamics were parameterized for five
major subarctic, arctic, and boreal ecosystem types:
upland tundra, black spruce forest, white spruce forest,
deciduous forest, and grassland-steppe. The most
recent version of the model instituted successional
dynamics for distinct shrub and graminoid tundra
vegetation classes and refined the migration dynamics
by which the graminoid and shrub tundra vegetation
classes transition to white spruce forest (Breen et al.
2013). Successional dynamics for shrub and grami-
noid tundra are probabilistic and influenced by fire
history and climate (Breen et al. 2013; Rupp et al.
2015). Burn severity is also a factor in the model,
influencing colonization and survival of white spruce
in tundra vegetation classes. There are five classes of
fire severity in the model: unburned (class 0), low burn
severity (class 1), moderate burn severity (class 2),
high crown severity with low surface severity (class
3), and high crown severity with high surface severity
(class 4). Low, moderate, and high crown severity with
low surface severity result in self-replacing succes-
sional trajectories for graminoid and shrub tundra;
whereas, high crown severity and surface severity fires
in shrub tundra can result in the transition from shrub
to graminoid tundra (Breen et al. 2013; Rupp et al.
2015). The likelihood of a high crown and surface
severity fire occurring decreases from spruce forest to
tundra vegetation classes and is discussed further in
Breen et al. (2013). Wetland tundra does not burn and
is a static vegetation class.
Successional dynamics for the transition from shrub
or graminoid tundra to white spruce forest are influ-
enced by fire history and climate (Breen et al. 2013;
Rupp et al. 2015). As an overview, a transition to white
spruce forest by a 1 km
2
tundra cell occurs from
succession or colonization and infilling (Fig. 2). Col-
onization of tundra by spruce is a two-step process
consisting of seed dispersal and seedling establish-
ment. These processes are influenced by climate and
fire history, which affect seedling establishment and
growth conditions. Within a given time step, if a fire
occurs in a tundra cell trees established in that cell
survive based on fire burn severity: 100 % survival
(class 1), 50 % survival (classes 2 and 3) and 0 %
Landscape Ecol
123
survival (class 4). The next step in the model is to check
whether there is a white spruce forest seed source
within an adjacent cell resulting in a 1 km
2
neighbor-
hood (Rupp et al. 2015). If a spruce seed source is
present, the model then checks whether conditions are
favorable for germination and establishment that are
modeled as a single process. Germination and estab-
lishment, however, cannot occur independently of seed
dispersal so we do not account for a seed bank in the
model. We assume white spruce seed viability at its
latitudinal and altitudinal limits is limited to the year of
dispersal. Climate is suitable for establishment and
growth (the next step) when the Summer Warmth
Index, the sum of mean monthly temperatures[0"C, is
[31 "C, and the ten-year moving-average of July
temperature is [10 "C. If conditions continue to be
favorable for growth, then basal area (BA) increases in
the cell with time. The basal area growth function
follows a normal distribution between 10 and 20 "C,
centered on 15 "C, for July mean temperature with a
maximum accrual of 2 mm per year. Mean July
temperatures \10 and [20 "C result in no growth
accrual, as these July isotherms bound the northern and
southern limits of the boreal forest in North America
(Larsen 1980). When the basal area of a cell reaches the
reproductive threshold [BA =10 meters
2
/hectare (m
2
/
ha)], it may act as a seed source for nearby cells. When
spruce trees become abundant in the cell and reach a
basal area threshold (BA =20 m
2
/ha), the cell tran-
sitions to the white spruce vegetation class. In addition
to climate, fire severity influences the basal area
accumulation in a cell. If a moderate or severe fire does
occur, then the spruce basal area of the tundra cell is
reduced by 50 % (class 2, 3) or 100 % (class 4) for the
shrub or graminoid tundra vegetation classes. In this
paper we present our findings from tree migration
simulations that incorporate modifications to the
tundra-to-spruce state transition to include the effects
of mycorrhizal fungi on establishment and growth
(Fig. 2) and fire-effects on EMF inoculum potential
and plant-fungal interactions detailed below (Fig. 3).
This is the first time that mycorrhizal effects (Fig. 3)
have been incorporated into ALFRESCO simulations
of tundra afforestation (Fig. 2).
Fig. 1 Conceptual diagram of the processes affecting state
transitions in ALFRESCO 2.0. Arrows indicate causal relation-
ships (Breen et al. 2013; Gray et al. 2013; Rupp et al. 2015). We
focus on the effects of EMF inoculum potential on successful
colonization of tundra by white spruce seedlings as indicated by
red arrows
Landscape Ecol
123
Mycorrhizal submodel parameterization
Previous field and greenhouse studies in Alaska have
shown that fire alters fungal composition (Hewitt et al.
2013; Hewitt 2014), and post-fire fungal composition
influences seedling performance for boreal tree
species expected to migrate into tundra with a
warming climate (Hewitt 2014). Studies from other
biomes show that limited availability of EMF, i.e.
lower EMF inoculum potential, reduced seedling
establishment, growth, and performance beyond the
forest edge (Perry et al. 1982,1987,1989; Dickie and
Reich 2005; Nunez et al. 2009). Collectively, this
suggests that seedling establishment, growth, and
migration potential may be limited at and beyond
treeline where pre-fire EMF inoculum potentials may
be low. In addition, increased wildfire occurrence and
severity may exacerbate these limiting effects of plant-
fungal interactions on tundra afforestation. We and
others have also observed that boreal seedlings can
share compatible fungi with EMF tundra shrubs
(Reithmeier and Kernaghan 2013; Hewitt 2014).
These findings together suggest that seedling estab-
lishment and growth, the most critical step in forest
migration, at and beyond current treeline may be
constrained by reduced EMF availability due to
variability in EMF host-plant densities across different
vegetation classes or wildfire effects.
To investigate the regional implications of land-
scape variability in EMF to land cover change we
developed four model parameterizations that represent
variation in EMF landscape inoculum potential and its
effects on seedling establishment phase and/or early
growth (see Table 1for description of model param-
eterization naming structure). For one of the model
parameterizations seedling establishment and growth
were not constrained by EMF (unconstrained inocu-
lum, UI). In the UI parameterization every forest and
tundra cell in the model was considered to have an
optimum inoculum potential and seedling establish-
ment and growth was unaffected by EMF. In contrast
to the UI parameterization, three EMF parameteriza-
tions represented variability in inoculum potential
across the model domain by deriving EMF inoculum
Fig. 2 Conceptual diagram of the processes affecting state
transitions from tundra to spruce in ALFRESCO 2.0. Arrows
indicate the progression from one step in the transition process
to the next step. The pathway to the right represents infilling of a
1 km
2
cell as long as there are spruce trees in the cell and
conditions are favorable. This is indicated by the different
hexagonal polygon. Figure modified from work by the
ALFRESCO 2.0 Team (Breen et al. 2013; Gray et al. 2013) to
emphasize (in red) the focus of this modeling effort, mycor-
rhizal effects on tree migration. Red indicates EMF (ectomy-
corrhizal fungi) inoculum potential modifiers to seedling
establishment and growth
Landscape Ecol
123
potentials from the pre-fire vegetation class of each
cell (e.g. highest EMF availability in forest, lowest in
graminoid tundra) and the fire history of each cell
(Fig. 3; Table 2). For these three EMF parameteriza-
tions, successional dynamics were constrained when
EMF availability, inoculum potential, was limited
Fig. 3 Conceptual diagram of the influence of mycorrhizal
fungi on tree seedling establishment and growth. Red indicates
EMF (ectomycorrhizal fungi) inoculum potential modifiers to
seedling establishment and growth. The arrows show the
progression from addressing ecological factors for one life
history stage to the next. Fire-severity classes are as follows:
0=unburned, 1 =low burn severity, 2 =moderate burn
severity, 3 =high crown severity with low surface severity
(LSS), and 4 =high crown severity with high surface severity
(HSS). Tundra classes are based on a highly modified output
originating from the North American Land Change Monitoring
System. We assigned inoculum potential scores to the tundra
classes based on shrub densities reported by Viereck et al.
(1992)
Table 1 Parameterizations of seedling establishment and growth with variation in the importance of limited EMF inoculum
potential on seedling performance
Abbreviated
name
Name Description
UI Unconstrained inoculum potential EMF inoculum potential does not vary across the landscape and is not
correlated with seedling establishment and growth
CI-E Constrained inoculum potential—
establishment
EMF inoculum potential varies across the landscape and the inoculum
potential is correlated with successful seedling establishment
CI-E ?STG Constrained inoculum potential—
establishment and short-term seedling
growth
EMF inoculum potential varies across the landscape and the inoculum
potential is correlated with successful seedling establishment and
short-term growth of seedlings. We define short term growth as the
accrual of seedling biomass in a 1 km
2
cell up until the cell has
reached a basal area of 1 m
2
/ha
CI-
E?LTG
Constrained inoculum potential—
establishment and long-term seedling
growth
EMF inoculum potential varies across the landscape and the inoculum
potential is correlated with successful seedling establishment and
long-term growth of seedlings. We define long-term growth of
seedlings as the accrual of seedling biomass in a 1 km
2
cell up until
the cell has reached a basal area of 5 m
2
/ha
Landscape Ecol
123
(Tables 1,2; Fig. 3). To elaborate, the shrub and
graminoid vegetation classifications in ALFRESCO
correspond to vegetation communities that differ in
their densities of plant functional types (tall and dwarf
shrubs, graminoids, and forbs) (Table 2). For exam-
ple, shrub tundra has a dominant shrub canopy with
less abundant forbs and graminoids, whereas grami-
noid tundra has dominant graminoids and lower
densities of shrubs and forbs. EMF inoculum poten-
tials for each vegetation class were derived from EMF
shrub densities described in the Viereck et al. (1992)
vegetation classification (see descriptions under II. C.
2 for shrub and graminoid tundra communities). To
illustrate, Viereck et al. (1992) described shrub tundra
as having up to 75 % cover by EMF shrubs (mainly
willows and dwarf birch). Because our previous
empirical work showed that EMF shrubs in Alaska
tundra host EMF compatible with boreal trees (Hewitt
2014) our EMF inoculum potential for shrub tundra
vegetation class is 75 % of the inoculum value for
white spruce forest vegetation class, which we
considered to have full inoculum potential. The
percent cover of shrub taxa capable of supporting
EMF compatible with boreal tree seedlings is reduced
threefold in graminoid tundra compared to shrub
tundra (Fig. 3; Table 2). As such, the white spruce
forest and tundra classes in simulations of treeline
dynamics are viewed as having a gradient of host plant
densities that support EMF compatible with boreal
tree seedlings (white spruce [shrub tun-
dra [graminoid tundra inoculum potential). The
assignment of these inoculum potentials includes
several assumptions about the mycobionts: (1) the
majority of EMF spores do not disperse far from intact
forest (Peay et al. 2010b,2012), and as such we
consider inoculum potential within a 1 km
2
cell as
independent of vegetation class and fire history of
neighboring cells, (2) EMF are generalists and EMF
taxa are compatible across multiple boreal tree and
tundra shrub taxa (Bent et al. 2011; Reithmeier and
Kernaghan 2013; Hewitt 2014), and (3) EMF promote
seedling performance (Hoeksema et al. 2010). These
broad assumptions are supported by our local findings
from empirical field and greenhouse studies after fire
in Alaska tundra: (1) post-fire tundra soils were not
strong sources of EMF inoculum beyond current
treeline (Hewitt 2014), (2) some EMF taxa can survive
fire on resprouting tundra shrub vegetation (Hewitt
et al. 2013) and colonize establishing boreal tree
seedlings (Hewitt 2014), and (3) post-fire mycobionts
are correlated with seedling nutrient status and
biomass (Hewitt 2014).
We parameterized the effects of fire severity on
EMF inoculum potential as part of the fire history
modifier to seedling establishment. Of the five burn
classes described above, we expect high severity fires,
classes three and four, to have a strong effect on
inoculum potential. Fire-severity effects on myco-
bionts are related to the degree to which host plants are
killed and the amount of soil combusted (Dahlberg
2002). The high severity fire classes with both low and
high surface severity reduce inoculum potential
through combustion of soil and killing or reduction
of EMF host plant densities. This is in contrast to
moderate and low severity fires, which we did not
parameterize to have reduced post-fire EMF inoculum
potentials. Despite a modest reduction in host plant
cover with moderate severity fires (e.g. reduced basal
area of spruce trees), there is low surface combustion
and no effect on tundra successional dynamics (Breen
Table 2 Derived inoculum scores from vegetation classifications
Model
vegetation class
Dominant plant functional types EMF inoculum
potential
Graminoid
tundra
Trees absent or sparse; dominant non-mycorrhizal or AM graminoids; lower densities of EMF
and ERM shrubs, up to 25 %; moss mat
0.25
Shrub tundra Trees absent or sparse; dominant tall and dwarf shrubs, up to 75 % (both EMF and ERM
hosts); lower densities of non-mycorrhizal or AM graminoids; sparse herbs
0.75
White spruce
forest
White spruce overstory; EMF and ERM shrub understory; feather moss mat 1.00
Inoculum potentials reflect dominance of cover by EMF hosts that support EMF compatible with boreal tree seedlings
AM arbuscular mycorrhizal fungi, EMF ectomycorrhizal fungi, ERM ericoid mycorrhizal fungi
Landscape Ecol
123
et al. 2013; Rupp et al. 2015) and thus we extrapolated
that reduction of inoculum potential is minimal. Low-
severity fires do not reduce spruce basal area or result
in tundra vegetation transitions and do not have
surface soil burning (Breen et al. 2013; Rupp et al.
2015). From our previous observations we assume that
low or no surface combustion and high potential of
resprouting host plants (e.g. after low and moderate
severity burning) (Hollingsworth et al. 2013) after fire
will maintain pre-fire inoculum potentials (Hewitt
et al. 2013; Hewitt 2014).
Three of four EMF parameterizations represent
seedling performance constrained by variability in
EMF across the model domain. These three EMF
parameterizations differ by whether EMF inoculum
potential affects seedling establishment and/or seed-
ling growth (Table 1). We limited mycorrhizal effects
to seedling stages. In the simulations seedling growth
is represented by basal area accrual. We used basal
area as a proxy for time and stand development based
on site-index curves for white spruce in interior Alaska
[i.e. 1 km
2
cell with BA =1m
2
/ha has small, very
young seedlings and 1\BA \5m
2
/ha has small,
young seedlings (Yarie and Cleve 1983)]. The three
different EMF parameterizations represent different
consequences for seedling performance when EMF
inoculum potential is reduced: (1) constraint on
seedling establishment (CI-E), (2) constraint on
seedling establishment and short-term growth (CI-
E?STG) for tundra cells with up to an accrued
spruce basal area of 1 m
2
/ha, and (3) constraint on
seedling establishment and long-term seedling growth
(CI-E ?LTG) for tundra cells with up to a basal area
of 5 m
2
/ha. The quantitative constraints of the EMF
parameterizations and establishment and growth are
based on the pre-fire inoculum potentials derived from
vegetation classes (white spruce =1.0; shrub tun-
dra =0.75; graminoid tundra =0.25 inoculum
potential) and fire-severity effects on inoculum poten-
tial (high crown severity with high surface severity
(class 4) =0.2; high crown severity with low surface
severity (class 3) =0.5; unburned (class 0), low burn
severity (class 1), moderate burn severity (class
2) =1.0 inoculum potential) described above
(Fig. 3). The seedling age and size at which EMF
inoculum potential most acutely affects seedling
performance is ambiguous in the empirical literature.
For example, the literature shows that EMF limitation
can hinder seedling establishment (Perry et al. 1982,
1989; Nunez et al. 2009), but if a seedling does
establish and has grown beyond sapling phase then it is
most likely that growth is not hindered at all by EMF
(Collier and Bidartondo 2009). As such a comparison
of the outputs of simulations for forest migration with
these four EMF parameterizations can serve as a
sensitivity analysis of EMF effects on forest migra-
tion. The comparison of these simulations represents
the full range of possible EMF-effects on stand
development from earliest effects (CI-E) to the latest
effects (CI-E ?LTG) based on what is known from
the empirical literature.
Climate and study domain
Simulations of forest migration and landscape
flammability across Alaska were driven by two
downscaled CMIP3 General Circulation Models
(ECHAM5 and CCCMA) that represent the best
performing models for the Alaska region (Rupp
et al. 2015) and a mid-range emissions scenario
(A1B) at a 1 km
2
resolution (Rupp et al. 2015).
Climate forcing from downscaled ECHAM5 outputs
represents a warm and dry scenario and the
CCCMA represents a scenario with climate condi-
tions warmer than the present but cooler and wetter
than ECHAM5.
In Alaska, the latitudinal treeline is the limit of
white spruce in the southern foothills of the Brooks
Range and Western Alaska. The latitudinal forest-
tundra ecotone is gradual, with spruce stand density
decreasing over a broad area, and in some localities
abrupt where physiographic features limit white
spruce growth (Viereck 1979). Our modeling exper-
iments focused on spruce expansion in the tundra
zone of Alaska (Fig. 4). The tundra zone in the
model includes arctic and subarctic lowland and
arctic-alpine tundra from the Brooks Range, North
Slope, and western regions of Alaska. The treeline
dynamics include two tundra classes (shrub and
graminoid) and one forest type (white spruce). We
compared land cover estimates from the beginning
of the century to those in 2100 (2000–2100). Our
general approach was to compare the outputs from
model simulations where three different EMF
parameterizations have been implemented (CI-E,
CI-E ?STG, CI-E ?LTG) to those that did not
include EMF-effects on seedling establishment and
stand development (UI) for both GCMs.
Landscape Ecol
123
Calibration
For vegetation transitions from tundra to spruce forest
model calibration was completed by comparing model
simulation results (1900–2005) to empirical studies
that estimated changes in vegetation cover. Model
simulation results obtained during the calibration
phase were a good fit for the historic period (Rupp
et al. 2015). Our target transitions were an increase in
white spruce forest cover of 5 % (Chapin et al. 2005;
Hinzman et al. 2005) and 30 % loss in graminoid
tundra due to shrub encroachment (Sturm et al. 2001;
Tape et al. 2006; Macias Fauria and Johnson 2008) in
the last century. These empirical observations were
our calibration transition targets for the historical
phase for all model parameterizations (the uncon-
strained parameterization and three EMF parameter-
izations) because our simulations should reflect forest
encroachment that has been observed in the field
where we assume a suite of abiotic and biotic factors,
including mycorrhizal fungi, influence successful
seedling establishment and growth.
Fire module calibration was accomplished using
historical fire activity records across the simulation
domain (Rupp et al. 2015). Simulations were evalu-
ated over the 1950–2009 time period, and parameters
were adjusted to approximate historical fire sizes,
frequency, and number of fires based on historical
observations. For metrics of landscape flammability,
model simulation results reasonably matched
observed (1950–2010) wildfire activity—both cumu-
lative area burned and the cumulative distribution of
individual fire size (Gustine et al. 2014).
Results
EMF effects on vegetation transitions
In the tundra zone of Alaska forest cover increased
over the simulation period for all parameterizations
(Figs. 5,6) by 8.8–18.2 % depending on model
parameterization (Table 1) and driving climate model
(Figs. 5,6). During the time frame from 2000 to 2100,
Fig. 4 Map of the model domain used in ALFRESCO simulations. We focus our analysis on forest expansion in the tundra zone, which
includes arctic and subarctic lowland and alpine tundra
Landscape Ecol
123
the simulations run with the constrained inoculum
establishment (CI-E) parameterization closely
matched those run with the unconstrained (UI)
parameterization and resulted in the greatest area of
forest that transitioned to forest from tundra compared
to the other parameterizations (Figs. 5,6). For exam-
ple the median percent change in forest cover from
2000 to 2100 for the CCCMA simulations was
18.25 % for the CI-E parameterization and 18.23 %
for UI parameterization. As the duration of the EMF
constraint increased (CI-E ?LTG [CI-E ?STG [
CI-E), afforestation rate declined (Fig. 5). Overall, the
largest reduction in forest expansion occurred when
simulations included effects of variability in inoculum
potential on establishment and long-term growth of
seedlings (CI-E ?LTG). This suggests that additive
mycorrhizal effects on establishment and growth
constrain forest expansion more than mycorrhizal
effects on establishment alone.
Irrespective of the model parameterization (uncon-
strained or EMF constrained), there was greater forest
expansion in the tundra domain for simulations driven
by the more moderate warming scenario, CCCMA,
than the ECHAM5 GCM. The CCCMA simulations
resulted in 6.7 % greater forest expansion by 2100
than the ECHAM5 simulations averaged across all
model parameterizations (Fig. 5). Simulations driven
by CCCMA had on average 6940 km
2
more forest
cover than ECHAM5 simulations by 2100 (Fig. 6).
The majority of forest expansion occurred in the
western Brooks Range (Fig. 7). Although there are
differences in the extent of tundra conversion among
simulations, the geographic domains where spruce
expansion occurred remain the same (Fig. 7). For
example, maps comparing UI and CI-E ?LTG sim-
ulations for both the CCCMA and ECHAM5 models
show greatest afforestation in northwestern Alaska
and a reduction in afforestation in these areas when
EMF effects are employed in simulations. Overall,
assessment of geographic patterns of afforestation
suggest that simulations that incorporate variability in
0.00
5.00
10.00
15.00
20.00
25.00
UI CI-E CI-E+STG CI-E+LTG
% change in forest cover
0
2
4
6
8
10
12
14
UI CI-E CI-E+STG CI-E+LTG
% change in forest cover
(a)
(b)
Fig. 5 Percent change in forest cover (km
2
±SD) for four
inoculum parameterizations across the tundra zone of the state
of Alaska from 2000 to 2100 averaged across 100 model
replicates: aCCCMA, bECHAM5
100000
105000
110000
115000
120000
125000
1950
1960
1970
1980
1990
2000
2010
2020
2030
2040
2050
2060
2070
2080
2090
2100
Simulation Year
UI-CCCMA
CI-E-CCCMA
CI-E+STG-CCCMA
CI-E+LTG-CCCMA
100000
105000
110000
115000
120000
1950
1960
1970
1980
1990
2000
2010
2020
2030
2040
2050
2060
2070
2080
2090
2100
Forest (km2) Forest (km2)
Simulation Year
UI-ECHAM5
CI-E-ECHAM5
CI-E+STG-ECHAM5
CI-E+LTG-ECHAM5
(a)
(b)
Fig. 6 Forest cover (km
2
) across the tundra zone of the state of
Alaska for simulations driven by four inoculum parameteriza-
tions: aCCCMA and bECHAM5 GCMs. Forest cover at each
time step is the average of 100 model replicate runs
Landscape Ecol
123
inoculum potential appear to influence the magnitude
of expansion within geographic localities instead of
altering which regions become forested compared
with simulations that do not incorporate EMF effects
(UI).
Maps differentiating spruce dispersal and establish-
ment (i.e., areas of spruce expansion that have not
transitioned to mature forest) from forest expansion [i.e.,
a tundra cell that has transitioned to a mature spruce
stand where the 1 km
2
cell has a basal area consistent
with a mature stand (BA [20 m
2
/ha)] illustrate that
EMF constraints on establishment and growth have
greater impact on afforestation than dispersal limitation
and EMF constraints on establishment alone (Fig. 7).
However, the magnitude of forest expansion differed
more based on forcing climate model than on inoculum
parameterization. Thus, the spatial extent of spruce
establishment and forest expansion (Fig. 7) mirror the
temporal trends in forest migration (Fig. 6) indicating
that climate and the associated fire regime is the primary
factor influencing the conversion of tundra to forest, and
secondarily limitation in EMF inoculum potential
constrains afforestation in simulations across the tundra
zone of the state.
Fig. 7 Comparison of the spatial patterns of landscape-level
forest expansion in the tundra zone of Alaska driven by four
inoculum parameterizations: aUI-CCCMA, bCI-E ?LTG-
CCMA, cUI-ECHAM5, and dCI-E ?LTG-ECHAM5. Effects
of variability in inoculum potential on establishment and long-
term growth of seedlings (CI-E ?LTG) had the largest
influence on forest expansion rates for simulations forced with
both climate models. Spruce establishment (orange pixels)
indicates where spruce has dispersed, but the tundra has not
transitioned to forest. Forest expansion (purple pixels) indicates
where spruce seed has dispersed during model simulations and
reached stand maturity. Expansion patterns displayed using the
best replicate simulation for the tundra zone (Breen et al. 2013)
Landscape Ecol
123
Indirect effects of EMF on landscape flammability
In parallel with our investigations of vegetation
transitions, we analyzed the indirect effects of myc-
orrhizal interactions on landscape flammability.
Because EMF parameterizations influence afforesta-
tion and thus fuel loads, these biotic interactions
indirectly affect landscape flammability. For each of
the GCMs there is fairly moderate variance in the
projected cumulative area burned for each of the four
model parameterizations (Fig. 8). Constraining EMF
parameterizations (CI-E, CI-E ?STG, and CI-
E?LTG) resulted in lower fire activity relative to
unconstrained parameterizations (UI) regardless of
driving GCM by 2100. The parameterizations with the
largest negative effect on afforestation (CI-E ?LTG)
had the most reduced landscape flammability. More
specifically, the simulations with unconstrained (UI)
parameterizations resulted in 11,974 km
2
(CCCMA)
and 18,045 km
2
(ECHAM5) more kilometers burned
than the CI-E ?LTG parameterization for the respec-
tive GCM. However, the cumulative area burned for
each of the eight simulations was more affected by the
abiotic factors, climate and fire, than the indirect effect
of variability in EMF on fuels (Fig. 8). For example,
cumulative area burned in the tundra was lower by an
average of 42,437 km
2
for simulations using CCCMA
than the warmer and drier ECHAM5 model, which is 2
or 3.5 times greater than the difference between UI and
CI-E ?LTG parameterizations driven by CCCMA
(11,974 km
2
) or ECHAM5 (18,045 km
2
), respectively
(Fig. 8).
We also observed that after high fire years there are
apparent negative feedbacks between lowered fuel
loads and subsequent periods of time with lower fire
activity. Due to the negative influence of limited EMF
on forest expansion, these negative feedbacks between
fire activity and fuels seem to be amplified in
simulations that are parameterized with constraints
on seedling establishment and growth due to low
EMF. As an example, notably high fire activity was
predicted in the 2001–2050 time period for simula-
tions forced with the ECHAM5 GCM. After the
highest peak in fire activity around 2035, variation in
cumulative area burned among model parameteriza-
tions is more apparent. This may reflect additive fire-
effects on plant-fungal interactions and forest expan-
sion, where fire reduces both seedling basal area and
EMF inoculum potential.
Discussion
The main objective of these modeling experiments
was to evaluate the landscape implications of our
understanding of how plant-mycorrhizal interactions
influence tundra-to-forest vegetation transitions. Here,
we found that, in the tundra zone of Alaska, abiotic
drivers, climate and associated fire activity primarily
influenced the magnitude of conversion to forest from
previously treeless areas. Yet, landscape patterns of
spruce expansion and forest development were con-
strained when we parameterized the model to limit
opportunities for positive EMF effects on tree estab-
lishment and growth in tundra. Mycorrhizal influence
on spruce migration consequently influenced the
magnitude of feedbacks between vegetation and fire
activity, providing insight into controls over vegeta-
tion response to disturbance and directional climate
change.
Variability in forest expansion in relation to the
model parameterization employed suggests that addi-
tive EMF-effects on establishment and biomass
accrual of seedlings and saplings, and not seedling
establishment alone, will most greatly influence
conversion of treeless tundra to forest on a land-
scape-scale. Historically, carbon limitation due to cold
temperatures has been proposed as the main control
0
20000
40000
60000
80000
100000
120000
140000
160000
180000
1950
1960
1970
1980
1990
2000
2010
2020
2030
2040
2050
2060
2070
2080
2090
2100
Cumulative area burned (km2)
Simulation year
UI-ECHAM5
CI-E-ECHAM5
CI-E+STG-ECHAM5
CI-E+LTG-ECHAM5
UI-CCCMA
CI-E-CCCMA
CI-E+STG-CCCMA
CI-E+LTG-CCCMA
Fig. 8 Mean cumulative area (km
2
) burned across the tundra
zone of the state of Alaska for simulations driven by four
inoculum parameterizations and ECHAM5 and CCCMA
GCMs. Cumulative area burned is the sum of the are burned
within a given year and all previous years starting from 1950
averaged across 100 model replicates
Landscape Ecol
123
over treeline seedling growth and treeline position
(Korner and Paulsen 2004; Hoch and Korner 2012).
More recently, nutrient limitation was put forth as a
key factor influencing treeline seedling (Sullivan and
Sveinbjornsson 2010) and stand development
(McNown and Sullivan 2013). Our simulations indi-
cate that stand development is sensitive to both
temperature and nutrient limitations and our findings
support a hierarchy of factors influencing treeline
change. For example, we observed the increased forest
cover over the next century for all simulations
indicating sensitivity to temperature, and the change
in forest cover was greater between driving GCMs
than among EMF parameterizations. Yet, within a
given climate scenario, forest expansion was con-
strained by a secondary factor, mycorrhizal effects-
likely associated with the influence of mycobionts on
both nutrient accrual and growth (Smith and Read
2008).
In a global synthesis, Harsch et al. (2009) inferred
that disturbance legacies, such as those from fire,
would likely not affect the probability of treeline
advance. Instead, the positive influence of climate on
afforestation would override any negative influence of
disturbance on initial recruitment. This hierarchy of
controls on treeline advance (climate followed by
disturbance legacies followed by biotic factors) is in
agreement with our model simulations, where climate
ultimately controls afforestation across the landscape,
but EMF factors influence the rate and/or magnitude of
reaching the climax state of the successional trajec-
tory, in this case spruce forest. Our findings are
consistent with several studies showing the impor-
tance of biotic factors such as seedbed quality and
herbivory (Cairns and Moen 2004; Munier et al. 2010)
as secondary filters and climate as the primary filter
affecting seedling establishment and growth beyond
treeline.
The magnitude of abiotic (such as climate and fire)
versus biotic (in our case EMF inoculum potential)
effects on afforestation might deviate from our model
outputs with the inclusion of EMF dispersal in
estimates of inoculum potential. Our best conceptual-
ization of estimated inoculum potential within a 1 km
2
cell in the model was based on pre-fire vegetation class
and fire history in that cell i.e. post-fire fungal survival
at the site (1 km
2
cell) excluding dispersal from
adjacent cells. This formulation does not fully capture
the biology of the plant-EMF symbiosis. However, we
represent a parsimonious conceptual model of EMF
inoculum potential reflective of rationale based on
several field experiments showing both local network-
ing of seedlings into intact hyphal networks (Horton
and van der Heijden 2008) and dispersal limitation at
distances less than 1 km (Peay et al. 2010b,2012)
influencing seedling performance. This representation
of inoculum potential at a coarse 1 km
2
–scale
excludes the potential positive ‘‘rescue effects’’ of
long-distance dispersal of spores on seedling growth
and the known negative effects of fungal limitation
that may constrain seedling growth at scales far below
1 km
2.
Thus, we suggest that our outputs likely result
in an underestimate of EMF effects on afforestation
despite the exclusion of dispersal in estimates of
inoculum potential.
Few studies outside of paleontological research
document the magnitude of forest vegetation shifts for
comparison to our estimates of vegetation transitions
from the model simulations. Paleontological studies
suggests that spruce-dominated latitudinal treeline is
currently at its farthest, northern Holocene-extent in
Alaska (Bigelow et al. 2003). Present-day den-
drochronological and satellite-based investigations
show increases in forest productivity at the forest-
tundra margin under favorable climate conditions
(Beck et al. 2011) coupled with observations of young
tree cohorts growing beyond the current forest limit
(Lloyd and Fastie 2003). Despite the retardation of
spruce migration due to effects of mycorrhizal limi-
tation on establishment and growth in our study,
simulations of afforestation were greater than
observed forest migration and tundra fire activity from
empirical studies within the 21st century. Chapin et al.
(2005) reported 2.3 % increase in forest cover over a
fifty-year time frame. From our study this is most
similar to the CI-E ?LTG-ECHAM5 simulations
with a conversion of 8.8 % tundra to spruce over a one
hundred year time frame. In strong juxtaposition to
this magnitude of tundra conversion, the CCCMA
simulations with the UI and the CI-E parameterization
were close to four-fold higher (*18 % tundra con-
version) and the ECHAM5 UI simulations over two-
fold higher (*11 % tundra conversion) than these
empirical estimates. Thus, the inclusion of some
restricting parameter on seedling establishment and
growth, whether EMF or otherwise, improved the
match between historic and projected estimates of
tundra conversion. Perhaps most importantly, future
Landscape Ecol
123
rates of treeline expansion may not reflect what has
been observed in empirical studies to date. With this in
mind the sensitivity analysis we conducted comparing
afforestation for all four model parameterizations for
each GCM yields a bracketed range of the extent of
forest migration that may occur in the next 100 years.
These high projections of state transitions are likely a
reasonable estimate given that currently the majority
of the model domain is within B2"C threshold of
summer temperatures that would likely promote forest
expansion and projections of future climate are on
average[2"C degrees C by 2100 (Scenarios Network
for Arctic and Alaska Planning 2015).
One of the main biophysical feedbacks discussed in
relation to treeline shifts and vegetation transitions in
the Arctic is the contribution to atmospheric heating.
Based on the estimates of changes in atmospheric
heating due to changes in forest cover reported in
Chapin et al. (2005), we calculated changes in
atmospheric heating for the changes in forest cover
that we found (Table 3). The most conservative
tundra-forest conversion was 0.9 % per decade, from
the CI-E ?LTG-ECHAM5 simulations. This resulted
in a *0.2 W m
2
/decade increase in atmospheric
heating. In contrast, the largest conversion from the
CCCMA simulation UI or CI-E parameterizations
(resulting in *1.8 % conversion per decade) would
result in a *0.4 W m
2
/decade increase in atmo-
spheric heating. While the contribution of vegetation
transitions to atmospheric heating are smaller than that
found for a reduced snow season due to climate
warming (*3.3 W m
2
/decade; (Chapin et al. 2005;
Euskirchen et al. 2009a,2009b), it is still important to
consider these tundra-forest conversions to come to a
more complete quantification of the full suite of
feedbacks to atmospheric heating.
The implications of vegetation transitions on long-
term atmospheric heating for a given climate scenario
is contingent on whether the influence of mycorrhizal
fungi retards the magnitude of transitions or prevents
afforestation. Our simulations are limited to a century-
long time frame ending in 2100. However, other
ALFRESCO simulations suggested substantial forest
expansion after 4000 years with a 2000 year time lag
when climate was ?9"C (Rupp et al. 2001). A
comparison of long-term simulations out to year 2300
that utilized climate scenarios with variation in
radiative forcing (decreased, constant, or increased
forcing), suggested that for long-term, directional
change to occur in response to climate warming (e.g.
sea ice decline), simulations required increased forc-
ing over the simulation period. In contrast, simulations
with decreased or constant forcing resulted in stasis or
recovery of a given ecological or physical state (Hezel
et al. 2014). For scenarios with increased forcing over
longer simulation time periods, it seems likely that our
mycorrhizal simulations represent a time lag in
seedling and sapling growth, with the climax spruce
state being reached over time. EMF inoculum would
likely increase across the tundra as vegetation types
with higher inoculum potential (forest and shrub
tundra) respond positively to climate and become
more dominant over the simulation time period (see
Breen et al. 2013). In contrast, climate scenarios with
constant or decreased radiative forcing from climate
might yield simulations with vegetation outputs that
would converge and the influence of mycorrhizal
fungi on vegetation processes would become more
pronounced. Given that warming is projected to
continue at high latitudes, the most likely extrapola-
tion from our simulations is that the forest will
continue to expand into tundra regions of Alaska,
Table 3 Calculated atmospheric heating (Watts m
-2
) for changes in forest cover from simulated tree migration based on estimates
of atmospheric heating reported in Chapin et al. (2005)
Parameterization % Afforestation Model: field W (m
-2
)
CCCMA ECHAM CCCMA ECHAM CCCMA ECHAM
UI 18.04 11.21 3.92 2.44 0.43 0.27
CI-E 18.18 11.33 3.95 2.46 0.43 0.27
CI-E ?STG 17.26 10.67 3.75 2.32 0.41 0.26
CI-E ?LTG 15.23 8.77 3.31 1.91 0.36 0.21
Model: field is a ratio of the percent change in forest cover from our model outputs compared to the field estimates reported in Chapin
et al. (2005) and used to calculate atmospheric heating
Landscape Ecol
123
influencing feedbacks between the biosphere and the
climate system.
Acknowledgments The Scenarios Network for Alaska and
Arctic Planning, the Alaska Climate Science Center, and the
Joint Fire Science Graduate Research Innovation Award
supported this research. We thank Shalane Frost for creating
Figs. 4and 7. The project described in this publication was
supported by Cooperative Agreement Number G10AC00588
from the United States Geological Survey. Its contents are solely
the responsibility of the authors and do not necessarily represent
the official views of the USGS.
Conflict of interest The authors declare that they have no
conflict of interest.
References
Beck PSA, Juday GP, Alix C, Barber VA, Winslow SE, Sousa EE,
Goetz SJ (2011)Changes in forest productivity acrossAlaska
consistent with biome shift. Ecol Lett 14(4):373–379
Bent E, Kiekel P, Brenton R, Taylor DL (2011) Root-associated
ectomycorrhizal fungi shared by various boreal forest
seedlings naturally regenerating after a fire in Interior
Alaska and correlation of different fungi with host growth
responses. Appl Environ Microbiol 77(10):3351–3359
Bever JD, Dickie IA, Facelli E, Faceli JM, Klironomos J, Moora
M, Zobel M (2010) Rooting theories of plant community
ecology in microbial interactions. Trends Ecol Evol
25(8):468–478
Bigelow NH, Brubaker LB, Edwards ME, Harrison SP, Prentice
IC, Anderson PM, Volkova VS (2003) Climate change and
Arctic ecosystems: 1. Vegetation changes north of 55
degrees N between the last glacial maximum, mid-Holo-
cene, and present. J Geophys Res 108(D19):8170
Breen AL, Bennett AP, Hewitt RE et al (2013) Tundra fire and
vegetation dynamics: simulating the effect of climate
change on fire regimes in Arctic ecosystems. Paper pre-
sented at the American Geophysical Union Fall Meeting,
San Fransisco, 9–13 December 2013
Cairney J, Bastias B (2007) Influences of fire on forest soil
fungal communities. Can J For Res 37:207–215
Cairns DM, Moen J (2004) Herbivory influences tree lines.
J Ecol 92(6):1019–1024
Chapin FS, Sturm M, Serreze MC, McFadden JP, Key JR, Lloyd
AH, Welker JM (2005) Role of land-surface changes in
Arctic summer warming. Science 310(5748):657–660
Collier FA, Bidartondo MI (2009) Waiting for fungi: the ecto-
mycorrhizal invasion of lowland heathlands. J Ecol
97(5):950–963
Dahlberg A (2002) Effects of fire on ectomycorrhizal fungi in
fennoscandian boreal forests. Silva Fenn 36(1):69–80
Dale VH, Joyce LA, McNulty S, Neilson RP (2001) Climate
change and forest disturbances. Bioscience 51(9):723–734
Dickie IA, Reich PB (2005) Ectomycorrhizal fungal commu-
nities at forest edges. J Ecol 93(2):244–255
Euskirchen ES, McGuire AD, Chapin FS, Yi S, Thompson CC
(2009a) Changes in vegetation in northern Alaska under
scenarios of climate change, 2003-2100: implications for
climate feedbacks. Ecol Appl 19(4):1022–1043
Euskirchen ES, McGuire AD, Rupp TS, Chapin FS, Walsh JE
(2009b) Projected changes in atmospheric heating due to
changes in fire disturbance and the snow season in the
western Arctic, 2003–2100. J Geophys Res Biogeosci
114(G4):G04022
Gardes M, Dahlberg A (1996) Mycorrhizal diversity in arctic
and alpine tundra: an open question. New Phytol 133(1):147–
157
Gray ST, Bennett AW, Bolton WR, Breen AL, Carman T (2013)
Using integrated ecosystem modeling to understand cli-
mate change. Alaska Park Sci 12(2):1–17
Gustine DD, Brinkman TJ, Lindgren MA, Schmidt JI, Rupp TS,
Adams LG (2014) Climate-driven effects of fire on winter
habitat for caribou in the Alaskan-Yukon Arctic. PLoS One
9(7):e100588
Harsch MA, Bader MY (2011) Treeline form—a potential key
to understanding treeline dynamics. Global Ecol Biogeogr
20(4):582–596
Harsch MA, Hulme PE, McGlone MS, Duncan RP (2009) Are
treelines advancing? A global meta-analysis of treeline
response to climate warming. Ecol Lett 12(10):1040–1049
Hewitt RE (2014) Fire-severity effects on plant-fungal interac-
tions: implications for Alaskan treeline dynamics in a
warming climate. PhD thesis, University of Alaska
Fairbanks
Hewitt RE, Bent E, Hollingsworth TN, Chapin FS, Taylor DL
(2013) Resilience of arctic mycorrhizal fungal communi-
ties after wildfire facilitated by resprouting shrubs. Eco-
science 20(3):296–310
Hezel PJ, Fichefet T, Massonnet F (2014) Modeled Arctic sea
ice evolution through 2300 in CMIP5 extended RCPs. The
Cryosphere 8(4):1195–1204
Hinzman LD, Bettez ND, Bolton WR, Chapin FS, Dyurgerov
MB, Fastie CL, Yoshikawa K (2005) Evidence and
implications of recent climate change in northern Alaska
and other arctic regions. Clim Change 72(3):251–298
Hobbie SE, Chapin FS (1998) An experimental test of limits to
tree establishment in Arctic tundra. J Ecol 86(3):449–461
Hoch G, Korner C (2012) Global patterns of mobile carbon
stores in trees at the high elevation tree line. Global Ecol
Biogeogr 21(8):861–871
Hoeksema JD, Chaudhary VB, Gehring CA, Johnson NC, Karst
J, Koide RT, Umbanhowar J (2010) A meta-analysis of
context-dependency in plant response to inoculation with
mycorrhizal fungi. Ecol Lett 13(3):394–407
Hollingsworth TN, Johnstone JF, Bernhardt EL, Chapin FS III
(2013) Fire severity filters regeneration traits to shape
community assembly in Alaska’s boreal forest. PLoS One
8(2):e56033
Horton TR, van der Heijden MGA (2008) The role of symbioses
in seedling establishment and survival. In: Leck MA,
Parker VT, Simpson RL (eds) Seedling ecology and evo-
lution. Cambridge University Press, Cambridge,
pp 189–213
Horton TR, Bruns TD, Parker VT (1999) Ectomycorrhizal fungi
associated with Arctostaphylos contribute to Pseudotsuga
menziesii establishment. Can J Bot 77(1):93–102
Hu FS, Higuera PE, Walsh JE, Chapman WL, Duffy PA, Bru-
baker LB, Chipman ML (2010) Tundra burning in Alaska:
Landscape Ecol
123
linkages to climatic change and sea ice retreat. J Geophys
Res Biogeosci 115:G04002. doi:10.1029/2009JG001270
Johnstone JF, Chapin FS (2003) Non-equilibrium succession
dynamics indicate continued northern migration of lodge-
pole pine. Glob Change Biol 9(10):1401
Johnstone JF, Chapin FS (2006) Effects of soil burn severity on
post-fire tree recruitment in boreal forest. Ecosystems
9:14–31
Johnstone JF, Hollingsworth TN, Chapin FS, Mack MC (2010)
Changes in fire regime break the legacy lock on succes-
sional trajectories in Alaskan boreal forest. Glob Change
Biol 16(4):1281–1295
Kelly R, Chipman ML, Higuera PE, Stefanova I, Brubaker LB,
Hu FS (2013) Recent burning of boreal forests exceeds fire
regime limits of the past 10,000 years. Proc Natl Acad Sci
USA 110(32):13055–13060
Korner C, Paulsen J (2004) A world-wide study of high altitude
treeline temperatures. J Biogeogr 31(5):713–732
Landhausser SM, Wein RW (1993) Postfire vegetation recovery
and tree establishment at the arctic treeline: climate-change-
vegetation-response hypotheses. J Ecol 81(4):665–672
Larsen JA (1980) The boreal ecosystem. Academic Press, New
York
Lloyd AH, Fastie CL (2003) Recent changes in treeline forest
distribution and structure in interior Alaska. Ecoscience
10(2):176–185
Macias Fauria M, Johnson EA (2008) Climate and wildfires in
the North American boreal forest. Philos Trans R Soc B
363(1501):2315–2327
McGuire AD, Sitch S, Clein JS, Dargaville R, Esser G, Foley J,
Heimann M (2001) Carbon balance of the terrestrial bio-
sphere in the twentieth century: analyses of CO2, climate
and land use effects with four process-based ecosystem
models. Glob Biogeochem Cycles 15(1):183–206
McNown RW, Sullivan PF (2013) Low photosynthesis of treeline
white spruce is associated with limited soil nitrogen avail-
ability in the Western Brooks Range. Alaska. Funct Ecol
27(3):672–683
Munier A, Hermanutz L, Jacobs J, Lewis K (2010) The inter-
acting effects of temperature, ground disturbance, and her-
bivory on seedling establishment: implications for treeline
advance with climate warming. Plant Ecol 210(1):19–30
Nara K (2006) Pioneer dwarf willow may facilitate tree suc-
cession by providing late colonizers with compatible
ectomycorrhizal fungi in a primary successional volcanic
desert. New Phytol 171(1):187–198
Nunez MA, Horton TR, Simberloff D (2009) Lack of below-
ground mutualisms hinders Pinaceae invasions. Ecology
90(9):2352–2359
Peay KG, Bidartondo MI, Elizabeth Arnold A (2010a) Not every
fungus is everywhere: scaling to the biogeography of
fungal–plant interactions across roots, shoots and ecosys-
tems. New Phytol 185(4):878–882
Peay KG, Garbelotto M, Bruns TD (2010b) Evidence of dis-
persal limitation in soil microorganisms: isolation reduces
species richness on mycorrhizal tree islands. Ecology
91(12):3631–3640
Peay KG, Schubert MG, Nguyen NH, Bruns TD (2012) Mea-
suring ectomycorrhizal fungal dispersal: macroecological
patterns driven by microscopic propagules. Mol Ecol
21(16):4122–4136
Perry D, Meyer M, Egeland D, Rose S, Pilz D (1982) Seedling
growth and mycorrhizal formation in clearcut and adjacent,
undisturbed soils in montana: a green-house bioassay. For
Ecol Manag 4(3):261–273
Perry DA, Molina R, Amaranthus MP (1987) Mycorrhizae,
mycorrhizospheres, and reforestation: current knowledge
and research needs. Can J For Res 17(8):929–940
Perry DA, Amaranthus MP, Borchers JG, Borchers SL, Brainerd
RE (1989) Bootstrapping in Ecosystems: internal interac-
tions largely determine productivity and stability in bio-
logical systems with strong positive feedback. Bioscience
39(4):230–237
Read DJ (1991) Mycorrhizas in ecosystems. Experientia
47(4):376–391
Reithmeier L, Kernaghan G (2013) Availability of ectomycor-
rhizal fungi to black spruce above the present treeline in
Eastern Labrador. PLoS One 8(10):e77527
Rupp TS, Starfield AM, Chapin FS (2000) A frame-based spa-
tially explicit model of subarctic vegetation response to
climatic change: comparison with a point model. Land-
scape Ecol 15(4):383–400
Rupp TS, Chapin FS, Starfield AM (2001) Modeling the influ-
ence of topographic barriers on treeline advance at the
forest-tundra ecotone in northwestern Alaska. Clim
Change 48(2–3):399–416
Rupp TS, Duffy P, Leonawicz M et al (2015) Climate scenarios,
land cover, and wildland fire. In: Zhu Z, McGuire AD (eds)
Baseline and projected future carbon storage and green-
house-gas fluxes in ecosystems of Alaska. U.S. Geological
Survey Professional Paper (In press)
Scenarios Network for Arctic and Alaska Planning (2015)
Average summer temperature data download. University
of Alaska. Available from http://www.snap.uaf.edu/tools/
data-downloads, Accessed 16 March 2015
Smith SE, Read DJ (2008) Mycorrhizal Symbiosis. Academic
Press, New York
Starfield AM, Chapin FS (1996) Model of transient changes in
arctic and boreal vegetation in response to climate and land
use change. Ecol Appl 6(3):842–864
Starfield A, Cumming D, Taylor R, Quadling M (1993) A frame-
based paradigm for dynamic ecosystem models. Ai Appl
7(2&3):1–13
Sturm M, Racine C, Tape K (2001) Increasing shrub abundance
in the Arctic. Nature 411(6837):546–547
Sullivan PF, Sveinbjornsson B (2010) Microtopographic control
of treeline advance in noatak national preserve, Northwest
Alaska. Ecosystems 13(2):275–285
Tape K, Sturm M, Racine C (2006) The evidence for shrub
expansion in Northern Alaska and the Pan-Arctic. Glob
Change Biol 12(4):686–702
Taylor DL, Herriott IC, Stone KE, McFarland JW, Booth MG,
Leigh MB (2010) Structure and resilience of fungal com-
munities in Alaskan boreal forest soils. Can J For Res
40(7):1288–1301
Turner MG (1989) Landscape ecology: the effect of pattern on
process. Annu Rev Ecol Syst 20:171–197
van der Heijden MGA, Horton TR (2009) Socialism in soil? The
importance of mycorrhizal fungal networks for facilitation
in natural ecosystems. J Ecol 97(6):1139–1150
Viereck LA (1979) Characteristics of treeline plant communi-
ties in Alaska. Ecography 2(4):228–238
Landscape Ecol
123
Viereck LA, Dyrness CT, Batten AR, Wenzlick KJ (1992) The
Alaska vegetation classification. General Technical Report
PNW-GTR-286 U.S. Department of Agriculture, Forest
Service, Pacific Northwest Research Station, Portland,
p 278
Yarie J, Cleve KV (1983) Biomass and productivity of white
spruce stands in interior Alaska. Can J For Res 13(5):
767–772
Landscape Ecol
123
A preview of this full-text is provided by Springer Nature.
Content available from Landscape Ecology
This content is subject to copyright. Terms and conditions apply.
