Repeated fires reduce plant diversity in low-elevation Wyoming big sagebrush ecosystems (1984 – 2014)

Abstract

Sagebrush is one of the most imperiled ecosystems in western North America, having lost about half of its original 62 million hectare extent. Annual grass invasions are known to be increasing wild-fire occurrence and burned area, but the lasting effects (greater than five years post-fire) that the resulting reburns have on these plant communities are unclear. We created a fire history atlas from 31 yr (1984-2014) of Landsat-derived fire data to sample along a fire frequency gradient (zero to three fires) in an area of northern Nevada that has experienced frequent fire in this time period. Thirty-two percent of our study area (13,000 km 2) burned in large fires (over 404 ha) at least once, 7% burned twice, and 2% burned three or more times. We collected plant abundance data at 28 plots (N = 7 per fire frequency), with an average time since fire of 17 yr. We examined fire's effect on plant diversity using species accumulation curves, alpha diversity (Shannon's dominance, Pielou's evenness, and number of species), and beta diversity (Whit-taker, Simpson, and Z indexes). For composition, we used non-metric multidimensional scaling. We then used PERMANOVA models to examine how disturbance history, temperature, precipitation, and aridity around the time of the fire affected subsequent community composition and diversity. One fire fundamentally changed community composition and reduced species richness, and each subsequent fire reduced richness further. Alpha diversity decreased after one fire. Beta diversity declined after the third fire. Cover of exotics was 10% higher in all burned plots, and native cover was 20% lower than in unburned plots, regardless of frequency. PERMANOVA models showed fire frequency and antecedent precipitation as the strongest predictors of beta diversity, while time since fire and vapor pressure deficit for the year of the fire were the strongest predictors of community composition. Given that a single fire has such a marked effect on species composition, and repeated fires reduce richness and beta diversity, we suggest that in lower elevation big sagebrush systems fire should be minimized as much as possible, perhaps even prescribed fire. Restoration efforts should be focused on timing with wet years on cooler, wetter sites.

Full text

Repeated fires reduce plant diversity in low-elevation Wyoming
big sagebrush ecosystems (1984–2014)
ADAM L. MAHOOD

AND JENNIFER K. BALCH
Department of Geography, University of Colorado Boulder, GUGG 110, 260 UCB, Boulder, Colorado 80309 USA
Citation: Mahood, A. L., and J. K. Balch. 2019. Repeated fires reduce plant diversity in low-elevation Wyoming big
sagebrush ecosystems (1984–2014). Ecosphere 10(2):e02591. 10.1002/ecs2.2591
Abstract. Sagebrush is one of the most imperiled ecosystems in western North America, having lost
about half of its original 62 million hectare extent. Annual grass invasions are known to be increasing wild-
fire occurrence and burned area, but the lasting effects (greater than five years post-fire) that the resulting
reburns have on these plant communities are unclear. We created a fire history atlas from 31 yr (1984–
2014) of Landsat-derived fire data to sample along a fire frequency gradient (zero to three fires) in an area
of northern Nevada that has experienced frequent fire in this time period. Thirty-two percent of our study
area (13,000 km
2
) burned in large fires (over 404 ha) at least once, 7% burned twice, and 2% burned three
or more times. We collected plant abundance data at 28 plots (N =7 per fire frequency), with an average
time since fire of 17 yr. We examined fire’s effect on plant diversity using species accumulation curves,
alpha diversity (Shannon’s dominance, Pielou’s evenness, and number of species), and beta diversity (Whit-
taker, Simpson, and Zindexes). For composition, we used non-metric multidimensional scaling. We then
used PERMANOVA models to examine how disturbance history, temperature, precipitation, and aridity
around the time of the fire affected subsequent community composition and diversity. One fire fundamen-
tally changed community composition and reduced species richness, and each subsequent fire reduced
richness further. Alpha diversity decreased after one fire. Beta diversity declined after the third fire. Cover
of exotics was 10% higher in all burned plots, and native cover was 20% lower than in unburned plots,
regardless of frequency. PERMANOVA models showed fire frequency and antecedent precipitation as the
strongest predictors of beta diversity, while time since fire and vapor pressure deficit for the year of the fire
were the strongest predictors of community composition. Given that a single fire has such a marked effect
on species composition, and repeated fires reduce richness and beta diversity, we suggest that in lower ele-
vation big sagebrush systems fire should be minimized as much as possible, perhaps even prescribed fire.
Restoration efforts should be focused on timing with wet years on cooler, wetter sites.
Key words: Artemisia tridentata ssp. wyomingensis; biodiversity; Bromus tectorum; cheatgrass; community composition;
fire; fire frequency; repeated fire; sagebrush.
Received 5 October 2018; revised 3 January 2019; accepted 11 January 2019. Corresponding Editor: Debra P. C. Peters.
Copyright: ©2019 The Authors. This is an open access article under the terms of the Creative Commons Attribution
License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
E-mail: adam.mahood@colorado.edu
INTRODUCTION
Wildfire activity has been increasing across the
western United States since the 1980s (Westerling
et al. 2006, Dennison et al. 2014, Westerling 2016,
Balch et al. 2017), and this is leading to concern
among land managers in the U.S. Great Basin
(Miller et al. 2013, Integrated Rangeland Fire
Management Strategy Actionable Science Plan
Team 2016, Chambers et al. 2017). This trend will
likely continue as rising temperatures and more
frequent drought events increase the probability
of fire (Krawchuk et al. 2009, Moritz et al. 2012,
Liu et al. 2013), and as these climatic factors com-
bine with increased human ignition pressure
(Balch et al. 2017) and land use change (Bowman
❖www.esajournals.org 1February 2019 ❖Volume 10(2) ❖Article e02591

et al. 2011) to increase the length of the fire sea-
son (Wotton and Flannigan 1993, Jolly et al.
2015). This increased fire activity is one con-
tributing factor to the loss of approximately half
of the area of sagebrush (Artemisia tridentata
Nutt.) shrubland communities, which once occu-
pied over 600,000 km
2
in the western United
States. Much of this land is now dominated by
cheatgrass (Bromus tectorum L.) (Bradley and
Mustard 2008), an introduced annual grass
(Davies 2011). This in turn is initiating a positive
feedback, wherein invading plants increase the
probability of fire, and increased fire activity
stimulates more annual grass invasion (D’Anto-
nio and Vitousek 1992, Brooks et al. 2004, Balch
et al. 2013). The result is a fire return interval that
has decreased from a historical range of 100–
342 yr for intact sagebrush (Baker 2006,
Bukowski and Baker 2013) to 78 yr in invaded
areas (Balch et al. 2013), to as low as 3–5yrin
cheatgrass-dominated areas in the Snake River
Plain (Whisenant 1990). This increase in fire activ-
ity results in more areas that are burned multiple
times, and the lasting effect this has on plant com-
munities’biodiversity and composition is rela-
tively unknown. There are relatively few studies
on the impacts of fire after more than 5 yr (but
see Beck et al. 2009, Reed-Dustin et al. 2016), and
fewer still that analyze the impacts of repeated
fires in the same location (Miller et al. 2013).
There are at least 40 vertebrate species of con-
servation concern associated with sagebrush
habitats (Rowland et al. 2006), including the
greater sage grouse (Centrocercus urophasianus).
Greater sage grouse depends on sagebrush for its
habitat and has been a management priority by
land managers (Chambers et al. 2017). Optimal
shrub cover for sage grouse is 15–25% with over
15% bunchgrasses and forbs (Beck et al. 2009).
Fire is one of the top 2 threats to the greater sage
grouse in the western part of its range (Brooks
et al. 2015), and the loss of sagebrush due to
wildfire has contributed strongly to its popula-
tion declines over the past 30 yr (Coates et al.
2016). Land management agencies have linked
fire management with long-term conservation
goals focused on sagebrush ecosystems and the
greater sage grouse (Chambers et al. 2017).
There is emerging consensus among research-
ers and land managers that lower elevation
Wyoming big sagebrush (A. tridentata ssp.
wyomingensis Beetle & Young) ecosystems are not
resilient to fire (Chambers et al. 2014) and should
be prevented from burning whenever possible,
while higher elevation Mountain big sagebrush
(A. tridentata ssp. vaseyana) ecosystems may still
recover naturally (Hanna and Fulgham 2015) or
with restoration by seeding (Knutson et al.
2014). Several authors have recommended
attempting to reduce the size and frequency of
wildfire, and stopping the use of prescribed fire
(Whisenant 1990, Baker 2006, Lesica et al. 2007,
Beck et al. 2009), while also reducing grazing
(Shinneman and Baker 2009, Ellsworth and
Kauffman 2013). Others have urged caution with
the use of prescribed fire (Davies et al. 2009,
Reed-Dustin et al. 2016, Shinneman and McIlroy
2016). There has been disagreement in the past
about the historical fire return interval for
Wyoming big sagebrush. It has been character-
ized as being every 35–100 yr (Schmidt et al.
2002), every 100–240 yr (Baker 2006), to every
171–342 yr (Bukowski and Baker 2013). This dis-
crepancy has important management implica-
tions, leading to disagreement as to which
stressors or disturbances (e.g., grazing, fire) need
to be increased or decreased in order to manage
for healthy sagebrush ecosystems. The lower
estimations imply the system is fire-dependent
and requires frequent burning in order to persist,
while the upper estimates suggest fire sensitivity.
Wyoming big sagebrush assemblages are gen-
erally agreed to be an endangered ecosystem and
fire, and the invasive plants that generally colo-
nize afterward are thought to be two major dri-
vers of declining biodiversity in this system
(Davies et al. 2011). Cover of introduced annual
grass species has been mostly observed to be
negatively related to species richness and native
diversity (Davies 2011, Gasch et al. 2013, Bansal
and Sheley 2016), but over a 45-yr period, Ander-
son and Inouye (2001) found that while intro-
duced annual grass cover was negatively
correlated with cover of native species, species
richness was unrelated. While fire is strongly cor-
related with annual grass cover in this system at
regional scales (Balch et al. 2013), it has also been
shown to be an unimportant predictor variable
for both exotic cover and species richness in east-
ern Washington (Mitchell et al. 2017).
Post-fire communities of introduced annual
grasses are affected by both fire frequency and
❖www.esajournals.org 2February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

time since fire. Cheatgrass cover can increase ini-
tially after fire, then stabilize above its pre-fire
cover after 2–5 yr (Reed-Dustin et al. 2016), but
positive linear relationships between time since
fire and cheatgrass cover have also been
observed (Shinneman and Baker 2009), as well as
areas where cheatgrass declined and was
replaced by perennial grasses (West and Yorks
2002, Hanna and Fulgham 2015). Pre-fire com-
munity composition might explain the inconsis-
tency in results. Cheatgrass can come to
dominate areas with fire-intolerant natives post-
fire, but in areas with pre-fire populations of fire-
tolerant species (e.g., Poa secunda J. Presl), these
species can regenerate following fire (Davies
et al. 2012, but see Bagchi et al. 2013).
Precipitation, temperature, and aridity affect
both the fire occurrence and the subsequent
recovery of plant communities. Unlike most
forested systems in the western United States,
burned area in Great Basin sagebrush systems is
best predicted by antecedent precipitation (Abat-
zoglou and Kolden 2013, Pilliod et al. 2017). Pre-
cipitation also drives the invasion of cheatgrass
into lower elevation sagebrush systems (Cham-
bers et al. 2007), which increases the probability
of fire for several years due to the persistence of
the litter it leaves behind (Pilliod et al. 2017).
Cheatgrass invasion increases the continuity of
fuels (Davies and Nafus 2013) and burned area
(Balch et al. 2013), thereby reducing the number
of unburned patches that provide the native seed
sources critical for recolonizing burned areas.
Unburned patches are essential for sagebrush
regeneration as almost every species in this
genus is a seed obligate and the seeds generally
fall no more than 30 m from the mother plant
(Meyer 1994). Once established, a sagebrush
seedling needs to be able to withstand drought
conditions in the summer to survive and be
recruited into the population (Meyer 1994).
Here, we explored how sagebrush community
composition and diversity responded to increas-
ing fire disturbance by constructing a fire history
atlas and sampling plant communities that
burned zero to three times between 1984 and
2014 in the Central Basin and Range ecoregion.
We constrained soil, ecological site type, eleva-
tion, and climate, and sampled blocks of plots
stratified along a gradient of zero to three fires.
Our first hypothesis was that community
composition would change drastically between
unburned and burned plots, but remain similar
between burned plots of different fire frequencies.
This was our expectation because in the Great
Basin there are vast areas of sagebrush which are
generally unburned in the last 30+yr, and burned
areas are almost always completely dominated by
cheatgrass, along with a handful of exotic forb
species and a native grass, P. secunda. These cheat-
grass-dominated areas all appear very similar,
regardless of fire frequency (Fig. 1). But we sus-
pected that there would be a signal on plant diver-
sity after multiple fires, due to selective pressure
against fire-intolerant plants. Thus, our second
hypothesis was that alpha diversity (the Shannon-
Weaver index, Pielou’s evenness, and the number
of species in a sampling unit), beta diversity (con-
tinuity or turnover of species between plots), and
the extrapolated species richness (with plots
pooled by fire frequency) would decrease with
increasing fire frequency. Our third hypothesis
was that cheatgrass abundance would have a neg-
ative relationship with plant diversity. Our fourth
hypothesis was that temperature, vapor pressure
deficit, and precipitation around the time of the
fire would exert a lasting influence over post-fire
community composition and diversity. This is
based on evidence that the effects of introducing
species at the beginning of secondary succession
can be long-lasting (Veen et al. 2018), and in this
system, the assemblage of species that are able to
successfully colonize an area after a fire depends
on their abilities to compete for moisture and tol-
erate drought (Meyer 1994).
METHODS
Study area
We conducted the study in a 13,000 km
2
region in northern Nevada (Fig. 2). The region
has hot, dry summers and cold, wet winters.
Annual precipitation averages 293 mm, falling
mostly from November to May. Mean tempera-
tures range from 21.8°C in July to 1.4°Cin
December (PRISM Climate Group 2016). The
region consists of mountain ranges that run
north–south, and the sagebrush ecosystems gen-
erally lie on the lower slopes of the mountains;
our sites ranged from 1272 to 1696 m in elevation
(median: 1458, SD: 99). From 1984 to 2014, 32%
(4096 km
2
) of the study area burned in large fires
❖www.esajournals.org 3February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

(over 500 acres) at least once, 7% burned twice,
and 2% burned three or more times (Appendix
S1: Table S1).
Site selection
We used a block sampling design, with each
block containing one site from each of four fire
frequencies (zero to three), and all fires occurring
at least five years before the study. We used
geospatial data representing ecosystem state fac-
tors (sensu Amundson and Jenny 1997) to design
a sampling scheme that constrained all other fac-
tors. We used the LANDFIRE (Rollins 2009) bio-
physical setting layer to eliminate all vegetation
types except big sagebrush shrubland. The
LANDFIRE data have 62–68% classification
accuracy for shrublands (Zhu et al. 2006). We
used soil data from the Natural Resource Conser-
vation Service to include only areas in the Loamy
8–10 precipitation zone (Soil Survey Staff, Natu-
ral Resources Conservation Service, United
States Department of Agriculture (USDA) 2016).
We chose this particular zone simply because it
was the most common type in our study area
within the big sagebrush shrubland biophysical
setting. We used the Land Treatment Digital
Library to exclude areas that had undergone
intensive restoration activities (Pilliod and Welty
2013). Excluding private and military land, and
areas more than 5 miles from a road eliminated
impractical plot locations and held human influ-
ence somewhat constant.
Fig. 1. Plot photographs taken from study block 4 in King’s River Valley, west of Orovada, Nevada. We show
these to illustrate the apparent similarity between plots with different fire frequencies, and why we thought spe-
cies composition would not change dramatically between one and three fires, while also hypothesizing that
diversity would decline.
❖www.esajournals.org 4February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

We accounted for additional, unknown distur-
bances such as grazing by using a block sam-
pling design and stratifying our statistical
analyses by these blocks. Long-term grazing data
were not available. Therefore, we assumed that
plots within blocks were close enough together
that they had experienced similar grazing pres-
sure. Additionally, we visually assessed the
impact of grazing on-site, aggregated what
records we could for the allotments in our study
(billed animal unit months (AUM) provided by
the Bureau of Land Management), and normal-
ized AUM by unit area and included these data
in our statistical modeling.
Once we constrained the area to a consistent
sampling space, we used Landsat-derived fire
data to stratify the space along a fire frequency
gradient. To generate fire history maps, we first
extracted only the values two to four (low, med-
ium, and high severity) from each yearly burn
severity mosaic from the Monitoring Trends in
Burn Severity (MTBS; Eidenshink et al. 2007)
project, as these were the values where one can
be reasonably certain that they actually burned.
Unburned patches and post-fire green-up, which
could be caused by a response to fire or an
unburned patch, were excluded. To generate fire
frequency maps, we reclassified each yearly layer
to a binary grid, and summed all 31 layers. To
avoid areas with less certain fire frequencies, we
then converted the MTBS fire perimeter polygons
to layers of fire frequency to extract only the grid
cells where the frequency from the polygons
matched the frequency from the reclassified ras-
ter grid. To generate last-year-burned maps, we
reclassified each severity mosaic (values two to
four) to the fire year, and calculated the maxi-
mum year for the entire time period for each
pixel. To eliminate areas that had burned more
recently than 2014, we masked pixels that
burned in 2015 according to the MODIS MCD64
burned area product (Giglio et al. 2009).
Kolden et al. (2015) have brought up several
shortcomings for the use of the MTBS burn
AB
C
Winnemucca
Inset (C)
Study Area (A)
NV
Fig. 2. The extent of the study area is shown in (A). The striping from the scanner line correction failure from
Landsat 7 is clearly visible, and those areas were avoided in our sampling. Darker shading indicates higher fire
frequency. The potential range is 0–5fires, although areas with more than three fires were extremely rare (0.2%
of total area). We sampled frequencies 0–3. The placement of the study area within the Central Basin and Range
ecoregion is shown in (B). A detail of one of the study blocks is represented in (C).
❖www.esajournals.org 5February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

severity mosaics, in particular inconsistent
development of class thresholds and a lack of
empirical relationships between the classified
values and ecological metrics. Because we only
used these data to get a more precise estimate of
fire occurrence (i.e., we used it to eliminate areas
of uncertainty) rather than using the severity
data as an independent variable for analysis, we
thought it sufficient to use these data in this
state. Another shortcoming that should be noted
is that there is no practical way for us to know
what these sites looked like before the earliest
fires in the fire record. The fact that our
unburned control plots were all mature sage-
brush is one piece of evidence suggesting these
sites were mature sagebrush pre-fire, but we
cannot be 100% certain, and this is a shortcom-
ing of all chronosequence studies (Walker et al.
2010).
We selected seven blocks in our sampling
space in accessible areas where there was a range
of fire frequencies and unburned areas for con-
trols within close proximity (0.5–10 km). Within
each block, we created spatially balanced ran-
dom points (Theobald et al. 2007) for each fire
frequency, and sampled one plot for each fire his-
tory class within the block. At each block, we
first sampled the unburned control plot to con-
firm that the area was indeed the correct vegeta-
tion type, and then sampled burned plots. After
navigating to the predetermined coordinates for
each plot, we first confirmed the physical charac-
teristics (soil type, lack of obvious restoration,
lack of obvious overgrazing) were within the
constraints of our sampling design. If a predeter-
mined point was not suitable (e.g., soil was too
rocky or sandy, an unburned control plot had
obviously burned, or it was the wrong ecological
site type), we referred to georeferenced PDFs of
our fire history atlas that we accessed with a sim-
ple application (Avenza Maps https://www.ave
nzamaps.com/) on a mobile device and located
nearby areas within the site that were suitable.
When a suitable area was found, we used a ran-
dom number generator to pick a random bearing
and a random distance, and navigated to the
new plot location.
We sampled 28 plots that fell along a gradient
of fire frequency (zero to three fires; N =seven
plots per frequency) and a range of times since
fire (4–31 yr; mean =17.6, SD =6.6; Fig. 2).
Because most of the fire effects research in this
system has been done within five years of a fire,
we aimed to have the time since fire of all of the
plots greater than or equal to 5 yr. We encoun-
tered 53 plant species—12 were introduced and
41 were native (Appendix S1: Table S2).
Plot establishment
We used GPS to navigate to predetermined
plot locations. Upon arrival, we established a
permanent marker at the southwest corner of the
plot. We recorded the slope, aspect, distance to
the nearest A. tridentata individual or other shrub
species, the topographic curvature of the site
(convex, concave, flat), evidence of ecological
restoration, grazing signs, and evidence of past
fires. We then delineated a 50 950 meter plot,
and placed pin flags at nine randomly deter-
mined 1 m
2
subplots within the plot with a mini-
mum spacing of 3 m. Pilliod and Arkle (2013)
found this sampling density sufficient for this
ecosystem, if supplemental methods are used to
estimate disparate functional groups like trees
and shrubs. Hence, we used the point-quarter
method as a supplement to estimate shrub cover
(see Pilliod and Arkle (2013) for detailed
methods).
Vegetation sampling
To explore how fire frequency influences com-
munity composition and diversity, we measured
the occurrence and abundance of all species. We
identified and recorded occupancy data for every
species within each subplot, and took a pho-
tograph from nadir with an Olympus Stylus
TG-870 digital camera to be analyzed later for
percent cover.
We used “Samplepoint”software (Booth et al.
2006) to analyze the digital photographs for per-
cent cover. We prepared photographs for analy-
sis by cropping them to the 1 91 m area of the
subplot. Then, we used Samplepoint to overlay a
regular grid of 100 points on each picture, and at
each point identified whether it was litter, bare
ground, rock, dung, or a plant. If it was a plant,
we identified it to species with the aid of the
occupancy data recorded at the plot. These data
were then converted to percent cover. If we
recorded a species as present within the subplot,
but it was missed by the photographic analysis,
we recorded it as 0.5% cover.
❖www.esajournals.org 6February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

Environmental data
Aspect was converted to folded aspect (folded
aspect =|180 |aspect 225||; McCune and
Keon 2002). This results in an approximation of
heat load ranging from zero (northeast) to 180
(southwest). Elevation was extracted from 10-m
resolution digital elevation models. The study
sites were situated among six grazing allotments.
To learn how climate before, during, and after
the fire event affected the subsequent community
composition and diversity, we extracted monthly
maximum vapor pressure deficit, maximum tem-
perature, and precipitation for the years before,
during, and after the most recent fire at each plot.
Maximum temperature and maximum vapor
pressure deficit were averaged for the entire year
before, during, and after, and precipitation was
averaged for the two winters (November–May)
prior and one after. We used monthly data pro-
vided by the PRISM Climate Group (2016) for all
climate variables. Variables used in modeling are
provided in Table 1. We also sampled soil C and
N (see Mahood (2017) for detailed methods).
Statistical analysis
Community composition and environmental
variables.—To analyze how fire frequency affects
community composition, we used non-metric
multidimensional scaling (NMDS). We ran a
rank correlation test for fire history gradients
against a matrix of relative cover of species per
plot to determine the best hierarchical clustering
method for creating a dissimilarity matrix. We
used this index for NMDS to examine how those
fire history characteristics affected the floristic
composition. To assess which species and envi-
ronmental variables had the most influence on
community composition, we added those vari-
ables to the ordinations using the “envfit”func-
tion from Vegan, with 9999 permutations and
stratified by the study block. Then, we grouped
species by their biogeographical origin (i.e.,
native or exotic), and used Tukey’s test to assess
how fire frequency influenced native cover, exo-
tic plant cover, and cheatgrass abundance.
Species richness, alpha diversity, and beta
diversity.—We created species accumulation
curves grouped by fire frequency to assess how
fire frequency affected species richness. This is dif-
ferent from alpha diversity in that the species
accumulation curve is estimating number of spe-
cies across all of the sites within each group with
each added plot, as opposed to simply calculating
a diversity index for each plot. We used the sam-
ple-based rarefaction method (Chiarucci et al.
2008, Oksanen et al. 2018, R Core Team 2016). We
used Tukey’s honestly significant difference test
(hereafter, Tukey’s test) to see whether different
fire frequencies influenced alpha diversity (the
Shannon-Weaver index, Pielou’s evenness, and
number of species per plot). There are several
ways to quantify beta diversity, most of which are
grouped into “measures of continuity”and “mea-
sures of gain and loss”(Koleff et al. 2003). We
used the “Z”index and Whittaker’soriginalbeta
diversity index for continuity measures, and
Simpson’s index (based on G. Gaylord Simpson’s
asymmetric index (Simpson 1943) and modified
by Lennon et al. (2001), not to be confused with
Edward H. Simpson’s index (1949)) for a measure
of gain and loss. To see how beta diversity dif-
fered between fire frequencies, we modeled the
homogeneity of dispersion of those matrices
(Anderson et al. 2006) and ran pairwise permuta-
tion tests (Legendre et al. 2011) on these models
Table 1. Variables used in PERMANOVA models.
Variable Abbreviation Source
Fire
Time since fire TSF MTBS
Fire frequency FF MTBS
Climate
Maximum vapor
pressure deficit
Year of fire vpdmax_during PRISM
Year before fire vpdmax_before PRISM
Year after fire vpdmax_after PRISM
Maximum temperature
Year of fire tmax_during PRISM
Year before fire tmax_pre PRISM
Year after fire tmax_after PRISM
Precipitation
November–May;
2 yr before fire
ppt_2pre PRISM
November–May;
1 yr before fire
ppt_1pre PRISM
November–May;
After fire
ppt_post PRISM
Other
Folded aspect Field
Measurements
Slope Field
Measurements
Elevation USGS
Animal unit months
per hectare
AUM_ha BLM
❖www.esajournals.org 7February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

with 9999 permutations, stratified by the study
blocks. To assess the influence of cheatgrass
abundance on alpha and beta diversity, we used
linear mixed models (Pinheiro et al. 2018) with
the study block as a random effect. We included
elevation as a fixed effect in addition to cheat-
grass due to its strong correlation with tempera-
ture and moisture availability, and ecosystem
resistance and resilience (Chambers et al. 2014).
We ensured that predictors had no multi-
collinearity using a variable inflation factor test
(Fox and Weisberg 2011), and used the partial
coefficient of determination (Jaeger et al. 2016) to
determine the cheatgrass component of the
model. To aid visualization, we removed the par-
tial effects of elevation from the dependent vari-
ables (Hohenstein and Kliegl 2018).
Modeling which fire and climate variables drive
post-fire composition and diversity.—To assess how
pre- and post-fire climate, along with soil and
other environmental variables (Table 1) affected
post-fire community composition and diversity,
we used permutational multivariate analysis of
variance (PERMANOVA). PERMANOVA uses a
dissimilarity matrix as the response variable and
columns from a separate data frame as the pre-
dictors. It makes the assumption that groups
being modeled have homogeneous dispersions.
If the test is run on groups with heterogeneous
dispersions, it is vulnerable to type 1 error
(Anderson and Walsh 2013). To account for this,
we built multivariate homogeneity of groups dis-
persions (MHGD) models on our community
clustering and beta diversity matrices grouped
by block, fire frequency, and burned vs unburned.
We then ran ANOVAs and Tukey’stestoneach
model, with Pvalues below 0.05 considered to be
an indication of heterogeneous dispersions. After
removing variables with multicollinearity, we
built PERMANOVA models with both commu-
nity clustering and beta diversity matrices using
an additive model-building process, with 9999
permutations, and stratifying the permutations by
the study blocks, with the aim of producing parsi-
monious models.
Code availability
Data and code to reproduce the analysis are
available at https://www.github.com/admahood/
ff_study and is on the dryad data repository
(doi:10.5061/dryad.520217j).
RESULTS
Community composition fundamentally changed
after one fire
The rank index test showed the Kulczynski
index to have the most consistent high scores
across gradients of fire history characteristics, so
we used this index for our hierarchical clustering
and NMDS analyses. Non-metric multidimen-
sional scaling (non-metric fit, R
2
=0.992, linear
fit, R
2
=0.972) showed seven unburned plots
clustered around high abundances of A. triden-
tata, and 18 burned plots clustered around B. tec-
torum (Fig. 3). Two thrice-burned plots were
dominated by exotic annual forbs (Sisymbrium
altissimum L. and Erodium cicutarium (L.) L’Her.
ex Aiton), and one was dominated by the native
perennial grass P. secunda (these are the three
thrice-burned plots outside of the “burned”
ellipse). The ordination showed a clear separa-
tion between burned and unburned plots, but
fire frequency was not significantly correlated
with the ordination, nor were any environmental
variables.
For the Tukey’s tests of exotic versus native
cover, there were differences between unburned
and burned plots (P<0.05) for both exotic
(increased by 10%) and native cover (decreased
by 20%), and no differences among the burned
plots (Fig. 4A, B). After dividing the mean cover
estimates into native and exotic life form groups
(annual and perennial graminoids and forbs, and
shrubs), we saw lower native shrub cover for
burned plots fire (24–3%), coupled with higher
annual grass cover (4–14%; Fig. 5).
Plant biodiversity decreased with each successive
fire
We found a decline in plant diversity at sites
that had burned more frequently. Species rich-
ness estimates declined as fire frequency
increased (Fig. 6; Appendix S1: Table S3). The
number of species and the Shannon-Weaver
index were higher in unburned plots, but the dif-
ferences were not significant, and Pielou’s even-
ness was not different between frequencies
(Fig. 4C–E). All three indexes of beta diversity
followed very similar patterns, so we only report
on Whittaker’s index here. It was not different
between zero and two fires, and lower for thrice-
burned plots (Fig. 4F), meaning that there is less
❖www.esajournals.org 8February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

dissimilarity within the group of thrice-burned
plots and more dissimilarity within the other
groupings.
Alpha diversity and evenness decreased with
cheatgrass abundance
Cheatgrass abundance had a negative relation-
ship with the Shannon-Weaver diversity (P0.05,
partial R
2
=0.65) and Pielou’s evenness (P0.05,
partial R
2
=0.51), a weak negative relationship
with the number of species (P<0.05, partial
R
2
=0.24), and no relationship to beta diversity
(P>0.5, partial R
2
=0.08; Fig. 7, Table 2). Eleva-
tion was important in all models except Pielou’s
evenness (Table 2).
Different climate and fire variables predict post-
fire composition and diversity
PERMANOVA models showed that fire history
and environmental factors influenced commu-
nity composition and beta diversity differently.
ANOVAs and Tukey’stestsonMHGDmodels
showed no heterogeneity in groups dispersions
for both beta diversity and hierarchical clustering
(P>0.05 for all models). Community composition
after fire was most affected by fire frequency, time
since fire, maximum vapor pressure deficit of the
year of the fire, and the interaction between fire
frequency and time since fire (Table 3, R
2
=0.55).
The relatively low amounts of variation accounted
for by the individual variables indicate these are
subtle effects. Beta diversity on the other hand was
influenced most by winter precipitation one and
two years prior to the fire, fire frequency, and
the interaction between winter precipitation one
year prior and max temperature for the year
after the fire (Table 4, R
2
=0.62). Here, the effect
was more pronounced, as more variation
accounted for by the three most statistically signif-
icant variables (fire frequency and precipitation
one and two winters prior to the fire).
DISCUSSION
The purpose of this study is to assess how
Wyoming big sagebrush plant communities
B
R
TE
TE
POS
E
A
R
T
R
W8
R
R
ELEL5
S
IAL
2
ERCI6
R
Unburned
Burned
−1.0
−0.5
0.0
0.5
1.0
−2 −1 0 1
NMDS1
NMDS2
Fire
Frequency
0
1
2
3
Fig. 3. Ordination plot of non-metric multidimensional scaling conducted on plant community data using Kul-
czynski hierarchical clustering. Ellipses represent the 95% confidence interval around plots grouped by whether
or not they had burned. Species significantly (P<0.05) correlated with the ordination are shown, with arrows
scaled by the strength of the correlation. Species are listed by their USDA plant codes. ARTRW8 is Artemisia tri-
dentata ssp. wyomingensis; POSE is Poa secunda; ELEL5 is Elymus elymoides; SIAL2 is Sisymbrium altissimum; BRTE
is Bromus tectorum; CETE5 is Ceratocephalum testiculatum; ERCI6 is Erodium cicutarium.
❖www.esajournals.org 9February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

respond to being burned repeatedly before
returning to their prior condition. The combina-
tion of a 32-yr fire history atlas and the use of the
RRQRR (Theobald et al. 2007) to randomly strat-
ify the sampling blocks over a large area pro-
vides broad-scale statistical inference for the
lower elevation (<1500 m) portion of the Wyom-
ing big sagebrush ecosystem. These lower
elevation sites generally experience higher tem-
peratures and lower soil moisture, and it is well
documented that they have lower resilience after
wildfires (Chambers et al. 2014). We did not
detect recovery of Wyoming big sagebrush at our
sites, and also found that while the cover of Bro-
mus tectorum does not change with successive
fires, the number of species in the species pool
does decrease and that biodiversity decreases
with cover of B. tectorum. The results of this
study may seem to conflict with other recent
studies documenting Wyoming big sagebrush
sagebrush recovery in the Great Basin (Ellsworth
et al. 2016, e.g., Shinneman and McIlroy 2016).
But all of the studies we are aware of showing
sagebrush recovery were conducted at cooler,
wetter sites, where Wyoming big sagebrush is
more resilient after fire (Chambers et al. 2014).
Coupling the 30+yr fire history atlas created
here with intensive field sampling offers a
unique opportunity to explore plant diversity
and composition changes in areas that have rela-
tively high fire frequencies, such as grass-domi-
nated or grass-invaded areas (Balch et al. 2013).
As annual grass invasions and their alterations to
fire regimes are a global phenomenon (D’Anto-
nio and Vitousek 1992, Brooks et al. 2004), this
type of study design will be useful for
D. Pielou’s evenness E. Number of species F. Beta diversity (Whittaker)
A. Native cover B. Exotic cover C. Shannon−weaver dominance
0123 0123 0123
0123 0123 0123
0.5
1.0
1.5
0.2
0.4
0.6
0
10
20
30
5
10
15
0
10
20
30
40
0.2
0.4
0.6
0.8
Fire frequency
Fig. 4. Alpha diversity (Shannon’s index of proportional abundance, Pielou’s index of evenness, and the num-
ber of species per plot), beta diversity (Whittaker’s index—the values are a unitless index of dissimilarity), and
native and exotic plant cover, all grouped by fire frequency. Shading indicates significantly different groups as
determined by Tukey’s test.
❖www.esajournals.org 10 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

understanding the consequences of changing fire
regimes in other regions. Additionally, new algo-
rithms are being developed that will lead to more
accurate and precise fire data products (Haw-
baker et al. 2015), leading to more nuanced fire
history atlases and thus more precise sampling
stratifications—especially now that burn severity
information can be easily incorporated (Eiden-
shink et al. 2007).
Community composition fundamentally changes
after one fire
In lower elevation A. tridentata ssp. wyomin-
gensis systems, our results show that one fire can
convert this shrub-dominated system to one
composed mainly of introduced annual grasses
and forbs, and we demonstrate that this new
state can persist for decades with little sign of
recovery to its prior condition. While almost all
of our burned plots were dominated by cheat-
grass, several thrice-burned plots were domi-
nated by P. secunda or exotic annual forbs (see
Fig. 3, where there are three plots that are out-
side the confidence envelope containing all other
burned plots). This corroborates previous work
showing that fire can push cheatgrass-invaded
grassland and shrubland communities into those
dominated by cheatgrass, P. secunda, and exotic
forbs, while uninvaded sites, or sites that are
invaded but still have significant bunchgrass
communities, can persist in a state of native
bunchgrasses and forbs (Davies et al. 2012, Reis-
ner et al. 2013, Condon and Pyke 2018). Other
studies have found that topography can be a
mediating factor, with native bunchgrasses more
likely to persist on steeper, more north-facing
slopes in the face of invasion and disturbance
(Rodhouse et al. 2014, Reed-Dustin et al. 2016).
One hypothesis that we were not able to test in
this study is that increasing fire frequency may
select for more fire-resilient plant functional
traits. More research is needed to investigate the
relationship between fire frequency and func-
tional traits. While it has been demonstrated that
B. tectorum establishes immediately post-fire and
can persist in the shorter term (Davies et al. 2012,
Hanna and Fulgham 2015), we show that this
0
10
20
30
0123
Fire frequency
Percent cover
Origin and life form
Exotic annual forb
Exotic annual grass
Exotic perennial forb
Exotic perennial grass
Native annual forb
Native perennial forb
Native perennial grass
Native shrub
Fig. 5. Percent cover of life form groups, grouped
by fire frequency. Of the two most dominant life form
groups, exotic annual grass is >99% cheatgrass, and
native shrub is >99% Wyoming big sagebrush.
10
20
30
246
Sites
Richness
Fire
Frequency
0
1
2
3
Fig. 6. Species accumulation curves for fire fre-
quency. Vertical lines represent the conditioned stan-
dard deviation around species richness and are jittered
for visibility.
❖www.esajournals.org 11 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

novel grass state can persist for long periods (i.e.,
>17 yr), corroborating recent work (Reed-Dustin
et al. 2016). If there was recovery, our study
design would have enabled us to detect it, as
Wyoming big sagebrush has been found to
recover from disturbance in as little as nine
(Wambolt et al. 2001) to 20 yr (Shinneman and
McIlroy 2016) following fire, and our fire history
atlas goes back 32 yr.
Biodiversity decreases with each subsequent fire
Here, we show that over a three decade period
repeated fires had long-lasting effects on commu-
nity composition and biodiversity in Wyoming
big sagebrush ecosystems. Species richness
declined with increasing fire frequency, but
measures of alpha and beta diversity decreased
after one and three fires, respectively (Fig. 4A, B).
Species accumulation curves demonstrated that
repeated fires are decreasing the overall pool of
species from which an individual patch might
draw from. So while there may not have been sig-
nificant differences in alpha diversity as fire fre-
quency increased, as the number of species each
plot can draw from decreased, this signal mani-
fested itself when beta diversity declined after
three fires.
We found negative relationships between
cheatgrass abundance and alpha and beta diver-
sity, as we hypothesized, but no relationship
between cheatgrass abundance and the number
of fires. Establishment and dominance of
C. Number of species D. Beta diversity (Whittaker)
A. Shannon−weaver dominance B. Pielou’s evenness
0 20 40 60 0 20 40 60
0 20 40 60 0 20 40 60
0.25
0.50
0.75
0.1
0.2
0.3
0.4
0.5
0.5
1.0
5
10
15
Cheatgrass cover
Elevation−adjusted values
Block
F00
F01
F02
F04
F05
F06
F10
Fig. 7. Scatter plots for a) Shannon-Weaver, b) Pielou’s evenness, c) number of species, and d) Whittaker’s beta
diversity as predicted by Bromus tectorum cover and elevation. Lines are predictions from linear mixed effects
models with study block as a random effect. The x-axis is cheatgrass cover, and the y-axis is the value of the index
with the effect of elevation removed (Hohenstein and Kliegl 2018).
❖www.esajournals.org 12 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

cheatgrass after fire are well documented (Whi-
senant 1990, Balch et al. 2013), and the relation-
ship between fire and species richness is clear
from this work. This implies that once an area is
invaded by cheatgrass, the competitive effects
from its increased abundance combine with its
effect on fire frequency to exclude species that
either cannot compete for moisture or cannot
survive fire. It should be noted that because we
selected sites that had burned at least three times
since 1984, we may have biased our results to be
applicable to only those areas that are susceptible
to initiating a grass–fire cycle.
Time since fire and vapor pressure deficit drive
community composition
PERMANOVA models showed that fire his-
tory and climate variables affect diversity and
Table 2. Results of linear mixed models testing the relationship between diversity indexes and cheatgrass abun-
dance, while accounting for elevation. Study block was the random effect. Partial coefficient of determination
was calculated from Jaeger et al. (2016).
Independent variable
Dependent variable
SW PE NS BD
B. tectorum cover 0.017*** (0.003) 0.008*** (0.002) 0.071*** (0.027) 0.002 (0.001)
Elevation 0.189*** (0.062) 0.046 (0.029) 1.700*** (0.642) 0.065*** (0.021)
Intercept 1.214*** (0.103) 0.652*** (0.052) 8.538*** (1.121) 0.288*** (0.037)
Partial R
2
,B. tectorum cover 0.65 0.51 0.24 0.08
Notes: BD, Beta Diversity (Whitaker); NS, Number of Species; PE, Pielou’s evenness; SW, Shannon-Weaver.
P<0.01.
Table 3. PERMANOVA results for fire history and environmental factors influencing post-fire community
composition.
Variable df SumsOfSqs MeanSqs F.Model R
2
P
TSF 1 0.1457 0.1457 2.0189 0.0713 0.0082
FF 1 0.2342 0.2342 3.2448 0.1146 0.0485
vpdmax_during 1 0.2243 0.2243 3.1081 0.1098 0.0016
tmax_during 1 0.0952 0.0952 1.3186 0.0466 0.1358
tmax_pre 1 0.0776 0.0776 1.0757 0.0380 0.1790
AUM_ha 1 0.2038 0.2038 2.8232 0.0997 0.2111
TSF:FF 1 0.1246 0.1246 1.7262 0.0610 0.0389
Residuals 13 0.9383 0.0722 0.4591
Total 20 2.0436 1.0000
Table 4. PERMANOVA results for fire history and environmental factors influencing post-fire beta diversity
(Whittaker’s index).
Variable df SumsOfSqs MeanSqs F.Model R
2
P
FF 1 0.2110 0.2110 2.4408 0.0783 0.0070
ppt_1pre 1 0.5236 0.5236 6.0581 0.1943 0.0014
tmax_after 1 0.1075 0.1075 1.2438 0.0399 0.0646
ppt_2pre 1 0.3993 0.3993 4.6199 0.1482 0.0134
TSF 1 0.1450 0.1450 1.6779 0.0538 0.2493
Folded aspect 1 0.1105 0.1105 1.2780 0.0410 0.4735
Elevation 1 0.0442 0.0442 0.5108 0.0164 0.9526
ppt_1pre:tmax_after 1 0.1162 0.1162 1.3447 0.0431 0.0255
Residuals 12 1.0371 0.0864 0.3849
Total 20 2.6943 1.0000
❖www.esajournals.org 13 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

community composition differently. Composi-
tion was found to be influenced by both fire fre-
quency and time since fire, and high vapor
pressure deficit the year of the fire. This suggests
that drought stress exerts a significant influence
on the particular plant species that will survive
and persist after a fire, and this effect can still be
detected decades after the fire burned. Shinne-
man and McIlroy (2016) also found that climatic
variables around the time of the fire influence the
eventual composition; namely, winter precipita-
tion the year after the fire was beneficial for sage-
brush recovery, but winter precipitation 2 yr
later had a negative effect. Elevation and recov-
ery have been shown to be positively related in
this system (Knutson et al. 2014), and most of the
studies showing fast recovery times were done at
higher elevations and latitudes (Wambolt et al.
2001, Hanna and Fulgham 2015, Ellsworth et al.
2016), in areas with long-term grazing exclusion
(Ellsworth et al. 2016), or on sites that were
specifically selected because their topographic
position was such that there was potential for
sagebrush recovery (Shinneman and McIlroy
2016). Here, we found that on low-elevation sites,
even after an average of 17 yr, post-fire sage-
brush cover was very low (<6%; also see Reed-
Dustin et al. 2016). These differences in recovery
rates (i.e., 9–20 yr at cooler sites vs no detectable
recovery at hotter sites) could be due to a slow-
ing down of recovery rates as the system loses
resilience with increasing drought stress at hotter
sites, while cooler sites have not yet experienced
sufficient drought stress to hamper recovery
(sensu van de Leemput et al. 2018).
Fire frequency and antecedent precipitation drive
beta diversity
Beta diversity was most heavily influenced by
fire frequency, precipitation for the two wet sea-
sons prior to the fire, and an interaction between
antecedent precipitation and maximum tempera-
ture for the year after the fire. Antecedent precip-
itation has been shown in other studies to be an
important predictor of fire occurrence and
burned area in this system (Abatzoglou and Kol-
den 2013, Balch et al. 2013). Since this is a fuel-
limited system, high precipitation increases fine
fuel loads and continuity (Davies and Nafus
2013), leading to higher fire probability, more
homogeneously burning fires, and larger extents.
Increased fine fuel loads could also be the driv-
ing factor behind decreasing diversity. Following
highly contiguous and extensive fires, there
would be fewer unburned patches as seed
sources, which are essential for the seed-obligate
sagebrush to reestablish quickly (Shinneman and
McIlroy 2016). In addition, Wyoming big sage-
brush is an opportunist in reproduction, setting
most of its seed in wet years (Meyer 1994) during
the short window in early spring when enough
water is available in the soil for plants to uptake
nutrients (Ryel et al. 2010, Schlaepfer et al. 2014).
So, in the years that Wyoming big sagebrush is
maximizing its expenditure on reproductive
resources, increased horizontal fuel continuity of
invasive annual grasses (Davies and Nafus 2013)
(1) increases the probability of burning and (2)
increases interspecific competition for resources
post-fire. This may result in a more homoge-
neous post-fire landscape populated mostly by
fire-tolerant plants.
Management implications
This work adds to the existing body of litera-
ture that suggests that in low-elevation
(<1700 m) Wyoming big sagebrush systems
wildfire should be minimized due to the negative
effects of single and repeated fires on community
composition and biodiversity. The reality is that
wildfire cannot be prevented, but fire suppres-
sion policies and practices could be crafted to
maximize the number and size of unburned
patches within burns to increase the probability
that Wyoming big sagebrush and other native
seed obligates recover post-fire. These results
also imply that prescribed burning is a risky
proposition with potentially disastrous conse-
quences for biodiversity and ecosystem structure
and function. However, we did not directly
assess the influence of prescribed fires in this
study. Prescribed fires typically are conducted at
a cooler time of year outside of or at the shoulder
of the fire season, and may have different ecolog-
ical effects due to the phenological stage plants
would be in at this different time of year, as well
as the lower burn severity that would be
expected due to cooler ambient air temperatures
and higher soil moisture. At a cooler, wetter site
where grazing has been excluded since 1994,
Ellsworth et al. (2016) detected the recovery of
sagebrush 17 yr after prescribed fires were
❖www.esajournals.org 14 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

conducted in late September 1997, which is the
natural end of the fire season at that location.
Two other studies at higher latitudes concluded
that prescribed burning to be an unwise action
even at those wetter sites. Beck et al. (2009) stud-
ied an area in southeast Idaho that was burned
in late August 1989 by prescribed fire 14 yr post-
fire for its utility in improving sage grouse habi-
tat. They recommended against prescribed fires
due to the lack of recovery of sagebrush. Wam-
bolt et al. (2001) found minimal benefit to the
herbaceous plant community at 13 sites that had
burned in prescribed fires in western Montana,
with little shrub recovery 6–15 yr after fire. Thus,
there is conflicting evidence on the use of pre-
scribed fires for management objectives even at
cooler wetter sites, providing less optimism for
the use of prescribed fires in the lower elevation
portion of the Wyoming big sagebrush ecosys-
tems studied here. Future research could focus
on comparing low-elevation Wyoming big sage-
brush sites that have been burned in prescribed
fires in the past paired with nearby areas that
burned in wildfires, with particular emphasis on
teasing out the effects of seasonality and burn
severity.
Our results from PERMANOVA modeling sug-
gest that the success of post-fire restoration efforts
will depend not only on elevation and topo-
graphic conditions (Arkle et al. 2014), but also the
climatic conditions that occur around the time of
the fire. This could mean that in a very dry year,
less money is spent on restoration efforts on low-
elevation sites, focusing instead on higher eleva-
tion sites and cooler aspects, and in wet years,
more funding is directed toward those more vul-
nerable low-elevation, southwest-facing sites.
Disagreement on the actual historical fire rota-
tion limits our ability to determine whether
Wyoming big sagebrush is fire-sensitive or fire-
resistant. However, this question may be irrele-
vant given the disruption and interaction
between invasive annual grasses and fires. We
demonstrate that when both fire and invasive
annual grasses operate in conjunction, sagebrush
is fire-sensitive. Moreover, we show that an alter-
nate exotic grass state can persist for 17 yr post-
fire even with only a single burn. This makes the
use of prescribed burning problematic, as the risk
of a fire-prone grassland establishing after a fire
likely outweighs the potential benefits of a
prescribed fire. Our results are specific to lower
elevation (<1700 m), dryer, hotter Wyoming big
sagebrush sites, and it remains to be explored
how sagebrush at higher elevations and latitudes
responds to increasing fire frequency, and how it
will respond under future climate change scenar-
ios. However, if temperatures continue to rise as
projected in this region (Garfin et al. 2014), those
areas may also become susceptible to a strong
grass–fire cycle. Overall, this effort demonstrates
that sagebrush communities are vulnerable to
repeated fires (Seipel et al. 2018), which should
be taken into account in land management deci-
sions (Chambers et al. 2017) that attempt to con-
serve or restore these valuable ecosystems, and
the threatened species that they harbor.
ACKNOWLEDGMENTS
We are grateful for the assistance of Nick Whittemore
and Kathleen Weimer for their assistance in the field
and in the laboratory. We thank Tom Veblen and Car-
son Farmer for comments on previous versions of the
manuscript.Weappreciatedtheconstructivecriticism
from three anonymous reviewers, which greatly
improved the paper. Thanks to Max Joseph for help
with the data analysis. We are also very grateful for the
support of the Nevada Bureau of Land Management
and the Central Nevada Interagency Dispatch Center.
This work was funded by the National Aeronautics and
Space Administration Terrestrial Ecology Program
under Award NNX14AJ14G and start-up funding from
the Department of Geography at CU Boulder. Publica-
tion of this chapter was funded by the University of
Colorado Boulder Libraries Open Access Fund.
LITERATURE CITED
Abatzoglou, J. T., and C. A. Kolden. 2013. Relation-
ships between climate and macroscale area burned
in the western United States. International Journal
of Wildland Fire 22:1003–1020.
Amundson, R., and H. Jenny. 1997. On a state factor
model of ecosystems. BioScience 47:536–543.
Anderson, M. J., K. E. Ellingsen, and B. H. McArdle.
2006. Multivariate dispersion as a measure of beta
diversity. Ecology Letters 9:683–693.
Anderson, J. E., and R. S. Inouye. 2001. Landscape-
scale changes in plant species abundance and
biodiversity of a sagebrush steppe over 45 years.
Ecological Monographs 71:531–556.
Anderson, M. J., and D. C. I. Walsh. 2013. PERMA-
NOVA, ANOSIM, and the Mantel test in the face of
❖www.esajournals.org 15 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

heterogeneous dispersions: What null hypothesis
are you testing? Ecological Monographs 83:557–
574.
Arkle, R., D. Pilliod, S. Hanser, M. L. Brooks, J. C.
Chambers, J. B. Grace, K. C. Knutson, D. A. Pyke, J.
L. Welty, and T. A. Wirth. 2014. Quantifying
restoration effectiveness using multi-scale habitat
models: implications for sage-grouse in the Great
Basin. Ecosphere 5:1–32.
Bagchi,S.,D.D.Briske,B.T.Bestelmeyer,andX.BenWu.
2013. Assessing resilience and state-transition mod-
els with historical records of cheatgrass Bromus
tectorum invasion in North American sagebrush-
steppe. Journal of Applied Ecology 50:1131–1141.
Baker, W. L. 2006. Fire and restoration of sagebrush
ecosystems. Wildlife Society Bulletin 34:177–185.
Balch, J. K., B. A. Bradley, J. T. Abatzoglou, R. C. Nagy,
E. J. Fusco, and A. L. Mahood. 2017. Human-
started wildfires expand the fire niche across the
United States. Proceedings of the National Acad-
emy of Sciences USA 114:2946–2951.
Balch, J. K., B. A. Bradley, C. M. D’Antonio, and J.
G
omez-Dans. 2013. Introduced annual grass
increases regional fire activity across the arid west-
ern USA (1980–2009). Global Change Biology
19:173–183.
Bansal, S., and R. L. Sheley. 2016. Annual grass inva-
sion in sagebrush steppe: the relative importance
of climate, soil properties and biotic interactions.
Oecologia 181:543–557.
Beck, J. L., J. W. Connelly, and K. P. Reese. 2009. Recov-
ery of greater sage-grouse habitat features in
Wyoming big sagebrush following prescribed fire.
Restoration Ecology 17:393–403.
Booth, D. T., S. E. Cox, and R. D. Berryman. 2006. Point
sampling digital imagery with ‘Samplepoint’. Envi-
ronmental Monitoring and Assessment 123:97–108.
Bowman, D. M. J. S., et al. 2011. The human dimension
of fire regimes on Earth. Journal of Biogeography
38:2223–2236.
Bradley, B. A., and J. F. Mustard. 2008. Comparison of
phenology trends by land cover class: a case study
in the Great Basin, USA. Global Change Biology
14:334–346.
Brooks, M. L., C. M. D. Antonio, D. M. Richardson, J.
B. Grace, J. E. Keeley, J. M. DiTomaso, R. J. Hobbs,
M. Pellant, and D. Pyke. 2004. Effects of invasive
alien plants on fire regimes. BioScience 54:677–688.
Brooks, M. L., J. R. Matchett, D. J. Shinneman and P. S.
Coates. 2015. Fire Patterns in the Range of the
Greater Sage-Grouse, 1984–2013—Implications for
Conservation and Management: U.S. Geological
Survey Open-File Report 2015-1167. Page 66.
Bukowski, B., and W. L. Baker. 2013. Historical fire
regimes, reconstructed from land-survey data, led
to complexity and fluctuation in sagebrush land-
scapes. Ecological Applications 23:546–564.
Chambers, J. C., B. A. Bradley, C. S. Brown, C. D’Anto-
nio, M. J. Germino, J. B. Grace, S. P. Hardegree, R.
F. Miller, and D. A. Pyke. 2014. Resilience to stress
and disturbance, and resistance to Bromus tectorum
L. invasion in cold desert shrublands of western
North America. Ecosystems 17:360–375.
Chambers, J. C., B. A. Roundy, R. R. Blank, S. E. Meyer,
and A. Whittaker. 2007. What makes Great Basin
sagebrush ecosystems invasible by Bromus tecto-
rum? Ecological Monographs 77:117–145.
Chambers, J. C., et al. 2017. Science Framework for
Conservation and Restoration of the Sagebrush
Biome: linking the Department of the Interior’s
Integrated Rangeland Fire Management Strategy
to Long-Term Strategic Conservation Actions. Part
1. Science basis and applications. Gen. Te:213.
Chiarucci, A., G. Bacaro, D. Rocchini, and L. Fattorini.
2008. Discovering and rediscovering the sample-
based rarefaction formula in the ecological litera-
ture. Community Ecology 9:121–123.
Coates, P. S., M. A. Ricca, B. G. Prochazka, M. L.
Brooks, K. E. Doherty, T. Kroger, E. J. Blomberg, C.
A. Hagen, and M. L. Casazza. 2016. Wildfire, cli-
mate, and invasive grass interactions negatively
impact an indicator species by reshaping sage-
brush ecosystems. Proceedings of the National
Academy of Sciences USA 113:12745–12750.
Condon, L. A., and D. A. Pyke. 2018. Fire and grazing
influence site resistance to Bromus tectorum through
their effects on shrub, bunchgrass and biocrust
communities in the Great Basin (USA). Ecosystems
21:1416–1431.
D’Antonio, C. M., and P. M. Vitousek. 1992. Biological
invasions by exotic grasses, the grass/fire cycle,
and global change. Annual Review of Ecological
Systems 23:63–87.
Davies, K. W. 2011. Plant community diversity and
native plant abundance decline with increasing
abundance of an exotic annual grass. Oecologia
167:481–491.
Davies, G. M., J. D. Bakker, E. Dettweiler-Robinson,
P. W. Dunwiddie, S. A. Hall, J. Downs, and J.
Evans. 2012. Trajectories of change in sagebrush
steppe vegetation communities in relation to multi-
ple wildfires. Ecological Applications 22:1562–
1577.
Davies, K. W., C. S. Boyd, J. L. Beck, J. D. Bates, T. J.
Svejcar, and M. A. Gregg. 2011. Saving the sage-
brush sea: an ecosystem conservation plan for big
sagebrush plant communities. Biological Conserva-
tion 144:2573–2584.
Davies, K. W., and A. M. Nafus. 2013. Exotic annual
grass invasion alters fuel amounts, continuity and
❖www.esajournals.org 16 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

moisture content. International Journal of Wildland
Fire 22:353–358.
Davies, K. W., T. J. Svejcar, and J. D. Bates. 2009. Inter-
action of historical and nonhistorical disturbances
maintains native plant communities. Ecological
Applications 19:1536–1545.
Dennison, P. E., S. C. Brewer, J. D. Arnold, and M. A.
Moritz. 2014. Large wildfire trends in the western
United States, 1984–2011. Geophysical Research
Letters 41:2928–2933.
Eidenshink, J., B. Schwind, K. Brewer, Z.-L. Zhu, B.
Quayle, and S. Howard. 2007. A project for Moni-
toring Trends in Burn Severity. Fire Ecology 3:3–21.
Ellsworth, L. M., and J. B. Kauffman. 2013. Seedbank
responses to spring and fall prescribed fire in
mountain big sagebrush ecosystems of differing
ecological condition at Lava Beds National Monu-
ment, California. Journal of Arid Environments
96:1–8.
Ellsworth, L. M., D. W. Wrobleski, J. B. Kauffman, and
S. A. Reis. 2016. Ecosystem resilience is evident
17 years after fire in Wyoming big sagebrush
ecosystems. Ecosphere 7:1–12.
Fox, J., and S. Weisberg. 2011. An R companion to
applied regression. Second edition. Sage, Thou-
sand Oaks, California, USA.
Garfin, G., G. Franco, H. Blanco, A. Comrie, P. Gonza-
lez, T. Piechota, R. Smyth and R. Waskom. 2014.
Southwest: The Third National Climate Assess-
ment. Pages 462–486 in J. M. Melillo, T. C. Rich-
mond, and G. W. Yohe, editors. Climate change
impacts in the United States: The Third National
Climate Assessment. U.S. Global Change Research
Program, Washington, D.C., USA.
Gasch, C. K., S. F. Enloe, P. D. Stahl, and S. E. Williams.
2013. An aboveground –belowground assessment
of ecosystem properties associated with exotic
annual brome invasion. Biology and Fertility of
Soils 49:919–928.
Giglio, L., T. Loboda, D. P. Roy, B. Quayle, and C. O.
Justice. 2009. An active-fire based burned area
mapping algorithm for the MODIS sensor. Remote
Sensing of Environment 113:408–420.
Hanna, S. K., and K. O. Fulgham. 2015. Post-fire vege-
tation dynamics of a sagebrush steppe community
change significantly over time. California Agricul-
ture 69:36–42.
Hawbaker, T. J., S. Stitt, Y.-J. Beal, G. Schmidt, J. Falgout,
B. Williams and J. Takacs. 2015. Provisional burned
area essential climate variable (BAECV) algorithm
description. United States Geological Survey.
Hohenstein, S. and R. Kliegl. 2018. Remef: remove par-
tial effects. R package version 1.0.6.9000. https://
github.com/hohenstein/remef
Integrated Rangeland Fire Management Strategy
Actionable Science Plan Team. 2016. The integrated
rangeland fire management strategy actionable
science plan. Page 128. U.S. Department of the Inte-
rior, Washington, D.C., USA.
Jaeger, B. C., L. J. Edwards, K. Das, and P. K. Sen. 2016.
An R
2
statistic for fixed effects in the generalized
linear mixed model. Journal of Applied Statistics
44:1086–1105.
Jolly, W. M., M. A. Cochrane, P. H. Freeborn, Z. A. Hol-
den, T. J. Brown, G. J. Williamson, and D. M. J. S.
Bowman. 2015. Climate-induced variations in glo-
bal wildfire danger from 1979 to 2013. Nature
Communications 6:7537.
Knutson, K. C., D. A. Pyke, T. A. Wirth, R. S. Arkle, D.
S. Pilliod, M. L. Brooks, J. C. Chambers, and J. B.
Grace. 2014. Long-term effects of seeding after
wildfire on vegetation in Great Basin shrubland
ecosystems. Journal of Applied Ecology 51:1414–
1424.
Kolden, C. A., A. M. S. Smith, and J. T. Abatzoglou.
2015. Limitations and utilisation of Monitoring
Trends in Burn Severity products for assessing
wildfire severity in the USA. International Journal
of Wildland Fire 24:1023–1028.
Koleff, P., K. J. Gaston, and J. J. Lennon. 2003. Measur-
ing beta diversity for presence-absence data. Jour-
nal of Animal Ecology 72:367–382.
Krawchuk, M. A., M. A. Moritz, M. A. Parisien, J. Van
Dorn, and K. Hayhoe. 2009. Global pyrogeogra-
phy: the current and future distribution of wildfire.
PLoS ONE 4:e5102.
Legendre, P., J. Oksanen, and C. J. F. ter Braak. 2011.
Testing the significance of canonical axes in redun-
dancy analysis. Methods in Ecology and Evolution
2:269–277.
Lennon, J. J., P. Koleff, J. Greenwood, and K. J. Gaston.
2001. The geographical structure of british bird dis-
tributions: diversity, spatial turnover and scale.
Journal of Animal Ecology 70:966–979.
Lesica, P., S. V. Cooper, and G. Kudray. 2007. Recovery
of big sagebrush following fire in southwest Mon-
tana. Rangeland Ecology & Management 60:261–
269.
Liu, Y., S. L. Goodrick, and J. A. Stanturf. 2013. Future
U.S. wildfire potential trends projected using a
dynamically downscaled climate change scenario.
Forest Ecology and Management 294:120–135.
Mahood, A. L. 2017. Long-term effects of repeated fires
on the diversity and composition of Great Basin
sagebrush plant communities. Dissertation,
University of Colorado Boulder, Boulder, Color-
ado, USA. https://scholar.colorado.edu/geog_grade
tds/111
❖www.esajournals.org 17 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

McCune, B., and D. Keon. 2002. Equations for poten-
tial annual direct incident radiation and heat load.
Journal of Vegetation Science 13:603–606.
Meyer, S. E. 1994. Germination and establishment ecol-
ogy of big sagebrush: implications for community
restoration. Pages 244–251 in Symposium on
management, ecology, and restoration of
lntermountain annual rangelands, boise, id, May
18–21, 1992.
Miller, R. F., J. C. Chambers, D. A. Pyke, F. B. Pierson,
C. J. Williams. 2013. A review of fire effects on veg-
etation and soils in the Great Basin Region:
response and ecological site characteristics. General
Technical Report RMRS-GTR-308. U.S. Department
of Agriculture, Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado, USA.
Mitchell, R. M., J. D. Bakker, J. B. Vincent, and G. M.
Davies. 2017. Relative importance of abiotic, biotic,
and disturbance drivers of plant community struc-
ture in the sagebrush steppe. Ecological Applica-
tions 27:756–768.
Moritz, M. A., M.-A. Parisien, E. Batllori, M. A. Kraw-
chuk, J. Van Dorn, D. J. Ganz, and K. Hayhoe.
2012. Climate change and disruptions to global fire
activity. Ecosphere 3:49.
Oksanen, J., et al. 2018. vegan: community ecology
package. R package version 2.5-3. https://CRAN.
R-project.org/package=vegan
Pilliod, D. S., and R. S. Arkle. 2013. Performance of
quantitative vegetation sampling methods across
gradients of cover in Great Basin plant communi-
ties. Rangeland Ecology & Management 66:634–
647.
Pilliod, D. S. and J. L. Welty. 2013. Land Treatment
Digital Library, Reston, Virginia, USA. https://ltdl.
wr.usgs.gov/
Pilliod, D. S., J. L. Welty, and R. S. Arkle. 2017. Refining
the cheatgrass-fire cycle in the Great Basin: precipi-
tation timing and fine fuel composition predict
wildfire trends. Ecology and Evolution 7:8126–
8151.
Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core
Team. 2018. nlme: linear and nonlinear mixed
effects models. R package version 3.1-137. https://
CRAN.R-project.org/package=nlme
PRISM Climate Group. 2016. PRISM gridded climate
data. Oregon State University, Corvallis, Oregon,
USA.
R Core Team. 2016. R: a language and environment for
statistical computing. R Foundation for Statistical
Computing, Vienna, Austria.
Reed-Dustin,C.M.,R.Mata-Gonz

alez, and T. J. Rodhouse.
2016. Long-term fire effects on native and invasive
grasses in protected area sagebrush steppe. Range-
land Ecology & Management 69:257–264.
Reisner, M. D., J. B. Grace, D. A. Pyke, and P. S.
Doescher. 2013. Conditions favouring Bromus tecto-
rum dominance of endangered sagebrush steppe
ecosystems. Journal of Applied Ecology 50:1039–
1049.
Rodhouse, T. J., K. M. Irvine, R. L. Sheley, B. S. Smith, S.
Hoh, D. M. Esposito, and R. Mata-Gonzalez. 2014.
Predicting foundation bunchgrass species abun-
dances: sagebrush steppe. Ecosphere 108:1–16.
Rollins, M. G. 2009. LANDFIRE: a nationally consis-
tent vegetation, wildland fire, and fuel assessment.
International Journal of Wildland Fire 18:235–249.
Rowland, M. M., M. J. Wisdom, L. H. Suring, and C.
W. Meinke. 2006. Greater sage-grouse as an
umbrella species for sagebrush-associated verte-
brates. Biological Conservation 129:323–335.
Ryel, R. J., A. J. Leffler, C. Ivans, M. S. Peek, and M. M.
Caldwell. 2010. Functional differences in water-use
patterns of contrasting life forms in Great Basin
steppelands. Vadose Zone Journal 9:548.
Schlaepfer, D. R., W. K. Lauenroth, and J. B. Bradford.
2014. Natural regeneration processes in big sage-
brush (Artemisia tridentata). Rangeland Ecology &
Management 67:344–357.
Schmidt, K. M., J. P. Menakis, C. C. Hardy, W. J. Hann
and D. L. Bunnell. 2002. Development of course-
scale spatial data for wildland fire and fuel man-
agement. General Technical Report RMRS-GTR-87.
U.S. Department of Agriculture, Forest Service,
Rocky Mountain Research Station, Fort Collins,
Colorado, USA.
Seipel, T., L. J. Rew, K. T. Taylor, B. D. Maxwell, and E.
A. Lehnhoff. 2018. Disturbance type influences
plant community resilience and resistance to Bro-
mus tectorum invasion in the sagebrush steppe.
Applied Vegetation Science 21:385–394.
Shinneman, D. J., and W. L. Baker. 2009. Environmen-
tal and climatic variables as potential drivers of
post-fire cover of cheatgrass (Bromus tectorum)in
seeded and unseeded semiarid ecosystems. Inter-
national Journal of Wildland Fire 18:191–202.
Shinneman, D. J., and S. K. McIlroy. 2016. Identifying
key climate and environmental factors affecting
rates of post-fire big sagebrush (Artemisia triden-
tata) recovery in the northern Columbia Basin,
USA. International Journal of Wildland Fire
25:933–945.
Simpson, G. G. 1943. Mammals and the nature of con-
tinents. American Journal of Science 241:1–31.
Simpson, E. H. 1949. Measurement of diversity. Nature
163:688.
Soil Survey Staff, Natural Resources Conservation
Service, United States Department of Agriculture
(USDA). 2016. Web Soil Survey. https://websoil
survey.sc.egov.usda.gov/
❖www.esajournals.org 18 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH

Theobald, D. M., D. L. Stevens, D. White, N. S. Urqu-
hart, A. R. Olsen, and J. B. Norman. 2007. Using
GIS to generate spatially balanced random survey
designs for natural resource applications. Environ-
mental Management 40:134–146.
van de Leemput, I. A., V. Dakos, M. Scheffer, and E. H.
van Nes. 2018. Slow recovery from local distur-
bances as an indicator for loss of ecosystem resili-
ence. Ecosystems 21:141–152.
Veen, G. F., W. H. van der Putten, and T. M. Bezemer. 2018.
Biodiversity-ecosystem functioning relationships
in a long-term non-weeded field experiment. Ecol-
ogy 99:1836–1846.
Walker, L. R., D. A. Wardle, R. D. Bardgett, and B. D.
Clarkson. 2010. The use of chronosequences in
studies of ecological succession and soil develop-
ment. Journal of Ecology 98:725–736.
Wambolt, C. L., K. S. Walhof, and M. R. Frisina. 2001.
Recovery of big sagebrush communities after burn-
ing in south-western Montana. Journal of environ-
mental management 61:243–252.
West, N. E., and T. P. Yorks. 2002. Vegetation responses
following wildfire on grazed and ungrazed sage-
brush semi-desert. Journal of Range Management
55:171–181.
Westerling, A. L. 2016. Increasing western US forest
wildfire activity: sensitivity to changes in the
timing of spring. Philosophical Transactions of the
Royal Society B: Biological Sciences. https://doi.
org/10.1098/rstb.2015.0178
Westerling, A. L., H. G. Hidalgo, D. R. Cayan, and T.
W. Swetnam. 2006. Warming and earlier spring
increase western U.S. forest wildfire activity.
Science 313:940–943.
Whisenant, S. G. 1990. Changing fire frequencies on
Idaho’s Snake River plains: ecological and
management implications. Pages 4–10 in E. D.
McArthur, E. M. Romney, S. Smith, and P. T. Tuel-
ler, editors. Proceedings of the Symposium on
Cheatgrass Invasion, Shrub Die-Off, and Other
Aspects of Shrub Biology and Management. Forest
Service General Technical Report INT-276. Inter-
mountain Research Station, Las Vegas, Nevada,
USA.
Wotton, B. M., and M. D. Flannigan. 1993. Length of
the fire season in a changing climate. The Forestry
Chronicle 69:187–192.
Zhu, Z., J. Vogelmann, D. Ohlen, J. Kost, X. Chen and
B. Tolk. 2006. Mapping existing vegetation compo-
sition and structure for the LANDFIRE prototype
project. Pages 197–215 in Gen. tech. rep. rmrs-gtr-
175. U.S. department of agriculture, forest service,
rocky mountain research station, Fort Collins, Col-
orado, USA.
SUPPORTING INFORMATION
Additional Supporting Information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/ecs2.
2591/full
❖www.esajournals.org 19 February 2019 ❖Volume 10(2) ❖Article e02591
MAHOOD AND BALCH