Ventenata dubia projected to expand in the western United States despite future novel conditions

Abstract

Distributions of both native and invasive species are expected to shift under future climate. Species distribution models (SDMs) are often used to explore future habitats, but sources of uncertainty including novel climate conditions may reduce the reliability of future projections. We explore the potential spread of the invasive annual grass ventenata (Ventenata dubia) in the western United States under both current and future climate scenarios using boosted regression tree models and 30 global climate models (GCMs). We quantify novel climate conditions, prediction variability arising from both the SDMs and GCMs, and the agreement among GCMs. Results demonstrate that currently suitable habitat is concentrated inside the invaded range of the northwest, but substantial habitat exists outside the invaded range in the Southern Rockies and southwestern US mountains. Future suitability projections vary greatly among GCMs, but GCMs commonly projected decreased suitability in the invaded range and increased suitability along higher elevations of interior mountainous areas. Climate novelty did not appear to undermine the prediction reliability in many cases where the climate–species relationship was fully represented by the occurrence data. GCM‐derived variability resulting from variation in future cool season precipitation and temperature seasonality was greatest in the Rocky Mountains. SDM‐derived variability was higher in currently suitable habitat, and few GCMs projections agreed that these areas would contain future suitable habitat. However, while prediction variability was high, many GCM projections agreed that parts of the Rocky, Wasatch, and Uinta Mountains would contain highly suitable habitat in the future. As disturbances in the interior mountains occur in coming decades, reducing some natural barriers to invasion, land managers, and conservationists will need to monitor for ventenata in post‐disturbance environments. Changes to invasion potential may not play out for several decades, but results related to current potential may have applications for early detection and rapid response planning.

Full text

ARTICLE
Climate Ecology
Ventenata dubia projected to expand in the western
United States despite future novel conditions
Ty C. Nietupski
1
| John B. Kim
2
| Claire M. Tortorelli
3
|
Rebecca Lemons
4
| Becky K. Kerns
5
1
USDA Forest Service, Geospatial
Technology and Applications Center, Fort
Collins, Colorado, USA
2
Western Wildland Environmental
Threats Assessment Center, USDA Forest
Service, Corvallis, Oregon, USA
3
Department of Plant Sciences, University
of California, Davis, Davis,
California, USA
4
Department of Forest Engineering,
Resources and Management, Oregon State
University, Corvallis, Oregon, USA
5
Pacific Northwest Research Station,
USDA Forest Service, Corvallis,
Oregon, USA
Correspondence
Ty C. Nietupski
Email: ty.nietupski@usda.gov
Present addresses
Claire M. Tortorelli, Pacific Northwest
Research Station, USDA Forest Service,
Corvallis, Oregon, USA; and
Rebecca Lemons, USDA Forest Service,
Klamath River Basin Landscape, Eureka,
California, USA.
Funding information
Oak Ridge Institute for Science and
Education (ORISE)
Handling Editor: Jennifer M. Fraterrigo
Abstract
Distributions of both native and invasive species are expected to shift under
future climate. Species distribution models (SDMs) are often used to explore
future habitats, but sources of uncertainty including novel climate conditions
may reduce the reliability of future projections. We explore the potential
spread of the invasive annual grass ventenata (Ventenata dubia) in the western
United States under both current and future climate scenarios using boosted
regression tree models and 30 global climate models (GCMs). We quantify
novel climate conditions, prediction variability arising from both the SDMs
and GCMs, and the agreement among GCMs. Results demonstrate that cur-
rently suitable habitat is concentrated inside the invaded range of the north-
west, but substantial habitat exists outside the invaded range in the Southern
Rockies and southwestern US mountains. Future suitability projections vary
greatly among GCMs, but GCMs commonly projected decreased suitability in
the invaded range and increased suitability along higher elevations of interior
mountainous areas. Climate novelty did not appear to undermine the predic-
tion reliability in many cases where the climate–species relationship was fully
represented by the occurrence data. GCM-derived variability resulting from
variation in future cool season precipitation and temperature seasonality was
greatest in the Rocky Mountains. SDM-derived variability was higher in cur-
rently suitable habitat, and few GCMs projections agreed that these areas
would contain future suitable habitat. However, while prediction variability
was high, many GCM projections agreed that parts of the Rocky, Wasatch,
and Uinta Mountains would contain highly suitable habitat in the future. As
disturbances in the interior mountains occur in coming decades, reducing
some natural barriers to invasion, land managers, and conservationists will
need to monitor for ventenata in post-disturbance environments. Changes to
invasion potential may not play out for several decades, but results related to
current potential may have applications for early detection and rapid response
planning.
Received: 15 December 2023 Revised: 3 May 2024 Accepted: 17 May 2024
DOI: 10.1002/ecs2.4979
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.
Published 2024. This article is a U.S. Government work and is in the public domain in the USA. Ecosphere published by Wiley Periodicals LLC on behalf of The Ecological
Society of America.
Ecosphere. 2024;15:e4979. https://onlinelibrary.wiley.com/r/ecs2 1of22
https://doi.org/10.1002/ecs2.4979

KEYWORDS
boosted regression trees, climate change, ecological niche modeling, global climate model,
invasive species, novel climate, species distribution modeling, uncertainty, Ventenata dubia
INTRODUCTION
Biological invasions are a leading cause of biodiversity
loss and have significant impacts on socioeconomic
systems worldwide (Bellard et al., 2016; Bradshaw
et al., 2016; Crystal-Ornelas et al., 2021; Shackleton
et al., 2019). The impact of invasions may become
increasingly severe as a result of changing climate, with
increased species translocations (Seebens et al., 2015,2018;
van Kleunen et al., 2015) and the reorganization of native
biotic communities (Bartley et al., 2019; Lurgi et al., 2012).
However, the success of invasive species varies by region
and ecosystem (Sorte et al., 2013), thereby reducing the
broadscale application of management strategies for some
invasions. When coupled with the increasing cost of inva-
sion management over time (Gaskin et al., 2021), manage-
ment practices like early detection and rapid response will
be essential in mitigating the negative impacts of invasive
species. The success of these practices will rely on informa-
tion that can adequately describe the influence of climate
change on a species’invasion potential (Funk et al., 2020;
Hulme, 2017).
Climate change is expected to impact the future distri-
bution of many plant species (Root et al., 2003; Walther
et al., 2002). Additionally, future wildfire activity is
expected to increase due to changes in temperature and
precipitation patterns (Abatzoglou & Williams, 2016;
Liu et al., 2010). The combined effect of changes to these
two drivers of species distribution may disproportionately
impact grass invasions. Invasive grasses can develop a
grass–fire cycle where grass invasion facilitates fire and
fire facilitates grass invasion (Brooks et al., 2004;
D’Antonio & Vitousek, 1992). As a result, understanding
the potential future distribution of invasive grasses
will be important to understanding potential synergistic
impacts of changing climate and wildfire disturbances on
native plant communities. For some species of invasive
grass, climate change may result in habitat reductions
(Gallagher et al., 2013), while habitat might shift or
increase for others (Abatzoglou & Kolden, 2011;
Bradley, 2009). Knowing where and when changes to
habitat are expected may provide enough warning for
managers to prevent the postfire proliferation of invasive
grasses in newly vulnerable ecosystems.
Species distribution models (SDMs), also known
as ecological niche models, have been used to predict
potential invasion across broad geographic domains
(Elith et al., 2010; Gallien et al., 2010; Peterson, 2003;
Srivastava et al., 2019; Thuiller et al., 2005). SDMs esti-
mate the relationship between a species and its environ-
ment, which can then be used to identify regions in space
or time where the environment is amenable for a species’
habitation (Guisan & Thuiller, 2005). While SDMs have
known limitations (Guillera-Arroita et al., 2015; Mainali
et al., 2015; Srivastava et al., 2019), they can provide a
reasonable assessment of the distribution of habitat or
potential range shifts when the characteristics of the
species, such as dispersal dynamics, species interactions,
or physiological constraints, are not yet fully understood
(Fordham et al., 2018). SDMs developed for invasive
species can be used to prioritize areas for heightened
management and prevention efforts, especially during
the early stages of invasion, when management efforts
may be more effective (Creutzburg et al., 2022).
The reliability of SDM predictions is affected by
uncertainty, which can arise from several aspects of the
modeling process in addition to the natural variability
of systems (Gould et al., 2014). An important source
of uncertainty comes from climate novelty (Elith
et al., 2010), where new climate does not resemble previ-
ously observed climate. To date, research on SDM appli-
cations in novel climate conditions has suggested many
modeling algorithms can produce unreliable predictions
in novel climate conditions (Mahony et al., 2017;
Owens et al., 2013). Therefore, it is commonly recom-
mended to assess climate novelty when predicting habitat
distributions to identify where model predictions may be
unreliable (Feng et al., 2019; Sofaer et al., 2019).
Uncertainty in climate suitability can also arise from
the global climate models (GCMs) used to project future
climate conditions (Gould et al., 2014). For instance,
projections of future habitat for cheatgrass (Bromus
tectorum) in the western United States showed up to a
135% difference in the future change of suitable habitat
depending on the GCM used to project future climate
(Bradley, 2009). However, it is uncommon for more than a
few GCM projections to be used for species distribution
modeling, which limits our understanding of the range and
variability of future invasion potential. Therefore, explicit
consideration of the differences and agreement among
many GCM-based predictions may provide a more wholistic
understanding of potential, future species distributions.
In this study, we use species distribution modeling to
examine the current and future invasion potential of a
2of22 NIETUPSKI ET AL.

rapidly spreading Eurasian annual grass, Ventenata dubia
(ventenata), across the western United States. Ventenata
is particularly problematic because it has the potential to
transform invaded areas into annual grasslands, resulting
in significant economic and ecological losses. Much
like other pervasive annual grasses in the western
United States, for example, cheatgrass, it increases fuel
loading and promotes fire in historically sparsely vege-
tated ecosystems (Tortorelli et al., 2023). Such altered
fire regimes can result in the loss of important plant spe-
cies (e.g., sagebrush) and wildlife habitat (D’Antonio &
Vitousek, 1992; Tortorelli et al., 2020). While ventenata
shares characteristics with other widely studied invasive
annual grasses, it appears to occupy a unique niche
and has heavily invaded areas previously resistant to
other annual grasses (Applestein & Germino, 2021;
Tortorelli et al., 2020,2022a). Despite invading pastures,
shrublands, grasslands, and dry forests, ventenata may
still be in an early stage of invasion (Jones et al., 2018).
The spread of ventenata into new landscapes and the
potential climate change impacts to its future distribution
are concerning given ventenata’s dispersal ability,
wide-ranging ecological tolerances, and capacity to cause
ecological and economic losses (Prather & Steele, 2009).
Understanding the potential for current and future inva-
sion of ventenata could help managers prioritize areas for
prevention and treatment while effectively utilizing lim-
ited resources.
Our study objectives are to: (1) develop an SDM that
explains both current and future climatically suitable
habitats; (2) examine ventenata’s relationship with cli-
mate, and (3) explore the forms of uncertainty that may
impact projections and our understanding of invasion
potential. While an SDM of ventenata’s current habitat
has been developed for the western United States
(Jarnevich et al., 2021), we employ a unique modeling
approach and explore climate relationships and uncer-
tainty in depth. In addition, this study provides the first
assessment of future ventenata invasion potential.
MATERIALS AND METHODS
Study area
The study area is the western conterminous United States,
all lands west of −104longitude (3 million km
2
;
Figure 1). The climate varies widely across the study area
and is Mediterranean in the coastal states, with mean
annual precipitation exceeding 2000 mm year
−1
in the
coastal northwest. Drier climates dominate the south and
much of the low-elevation interior. In the interior, signifi-
cant summer rainfall is delivered by convective storms.
Summer rainfall is particularly important in the southern
portion of the region, where the monsoon season brings
up to 50% of the annual rainfall (Higgins et al., 1997).
Nearly half (46%) of the western United States is covered
by shrublands, the majority of which are spread between
the Great Basin and the southwestern deserts (Figure 1;
Dewitz, 2019). Forests, mostly in mountainous areas, cover
22% of the western United States and the remaining third
is split between grassland or herbaceous cover (18%),
agriculture (7%), development (3%), and wetlands (1%).
Species data
Ventenata occurrence data were drawn from observa-
tional and experimental studies and publicly available
databases (Appendix S1: Table S1). Occurrence records
were cleaned and compiled, resulting in 2083 observa-
tions spanning the years 1957–2018. Most records were
found in the interior northwest but included all states
with recorded occurrence in the USDA plants database
(https://plants.udsa.gov), except for Wyoming.
We minimized the potential impacts of sampling
bias by spatially thinning the occurrence dataset using a
5 km minimum distance between occurrence points
(Boria et al., 2014; Phillips et al., 2009), resulting in
563 observations. To retain information potentially lost
during the thinning process, we created 100 different
thinned datasets of occurrence for use in model develop-
ment. We also weighted occurrence records based on
an observation’s proximity to the nearest major road
(protocols detailed in Appendix S1: Section 1.2; Statistics
Canada, 2015; U.S. Geological Survey, 2014) and the dis-
tance to the invasion origin.
Absence data were generated (i.e., pseudo-absence) to
control for nonequilibrium distribution issues, as the spe-
cies may still be expanding in extent or abundance in the
invaded ranges (Thuiller et al., 2005). The generation of
pseudo-absences can impact the generality and bias
of SDMs (Srivastava et al., 2019). However, sampling
bias in occurrence data has a greater impact to model bias
and generality than the method of pseudo-absence genera-
tion (Kramer-Schadt et al., 2013; Stokland et al., 2011). We
focused on methods to minimize bias in the occurrence
data and implemented pseudo-absence generation
methods that have been shown to improve SDMs, includ-
ing limiting the proximity of pseudo-absences and
restricting the total geographic extent of pseudo-absence
generation. We applied two constraints to generate
pseudo-absences. First, we defined the “invaded range”as
a 50-km buffer of the convex hull around all occurrence
points (Figure 1; Mainali et al., 2015). We (1) randomly
generated 100 replicates of 10,000 pseudo-absences within
ECOSPHERE 3of22

FIGURE 1 Ventenata (Ventenata dubia) occurrence and the diversity of vegetation, ecoregions, and land cover types in the western
United States. The white shaded polygon shows ventenata’s invaded range (i.e., the minimum convex hull around occurrence points
buffered by 50 km).
4of22 NIETUPSKI ET AL.

the invaded range, (Elith et al., 2010; Mainali et al., 2015)
and (2) excluded areas within 5 km of any individual occur-
rence (Barbet-Massin et al., 2012). The pseudo-absences
were combined with the thinned ventenata occurrence
points. Finally, we weighted individual observations to
account for the sample-size imbalance between the number
of occurrence points and pseudo-absence points, ensuring
that the sum of pseudo-absence weights equaled the sum of
occurrence weights (Barbet-Massin et al., 2012).
Environmental data
Current climate was represented with 30-year normals cal-
culated based on water-year (October 1982–September 2012)
at 800-m resolution (PRISM LT71; PRISM, 2012). From
these climate normals, we calculated 19 biocli-
matic variables (Hijmans et al., 2021; Nix, 1986) to char-
acterize extremes, annual trends, and seasonal trends in
long-term climate that could limit physiological functions
and, consequently, species distributions (Table 1). We
calculated five additional variables that characterize the
critical periods of the growing season to address physio-
logical controls on annual grass distributions (Gardner
et al., 2019). While fall climate can be important for
ventenata, germination requirements may not be met
until winter or spring periods (Wallace et al., 2015).
Therefore, we only included winter and spring in our def-
inition of the “cool season.”To account for potential vari-
ation in the timing of seasons at the end of the century,
we tested two different ways to define spring: (1) the
quarter after the coldest quarter, and (2) the first quarter
where the mean monthly temperature exceeds 5Cin
all three months (Linderholm, 2006). In addition, we
defined cool season precipitation as the total precipita-
tion in both the coldest quarter and the quarter immedi-
ately following the coldest quarter. Metrics related to
soils and topography were evaluated in the early phases
of the model development but had negligible influence
and were excluded from further consideration.
Future climate was represented by the NASA
Earth Exchange Downscaled Climate Projections dataset
(NEX-DCP30; Thrasher et al., 2013). We used climate pro-
jections from 30 GCMs for Representative Concentration
Pathway 8.5 (RCP8.5; van Vuuren et al., 2011). Further
details on these 30 GCMs are provided in Appendix S1:
Table S2. Climate projection data were accessed through
Google Earth Engine (Gorelick et al., 2017), where they
were processed to climate normals for the end of the
century (October 2069–September 2099). Previously
described climatic variables were calculated for each
GCM, resulting in gridded predictions of future climate
at an 800-m resolution.
Predictor selection and model
transferability
Within the current climate space, we identified highly
correlated variables (Pearson correlation threshold of
±0.85; Elith et al., 2010) and retained those with stronger
ecological importance for ventenata based on the litera-
ture and the winter-annual grass life cycle. To minimize
impacts to model spatial and temporal transferability, we
calculated the correlation structure difference between
(1) the invaded range and the western United States
and (2) the current and future climate scenarios in the
western United States (Appendix S1: Figures S1–S4).
TABLE 1 Climatic predictors explored for the species
distribution modeling of ventenata (Ventenata dubia).
Code Description Units
bio1 Annual mean temperature C
bio2 Mean diurnal range C
bio3 Isothermality C
bio4 Temperature seasonality C
bio5 Maximum temperature of warmest month C
bio6 Minimum temperature of coldest month C
bio7 Temperature annual range C
bio8 Mean temperature of wettest quarter C
bio9 Mean temperature of driest quarter C
bio10 Mean temperature of warmest quarter C
bio11 Mean temperature of coldest quarter C
bio12 Annual precipitation mm
bio13 Precipitation of wettest month mm
bio14 Precipitation of driest month mm
bio15 Precipitation seasonality mm
bio16 Precipitation of wettest quarter mm
bio17 Precipitation of driest quarter mm
bio18 Precipitation of warmest quarter mm
bio19 Precipitation of coldest quarter mm
csppt Cool season precipitation mm
qacqppt Precipitation of quarter after coldest
quarter
mm
qacqtavg Mean temperature of quarter after coldest
quarter
C
sosppt Precipitation of start of season quarter mm
sostavg Mean temperature of start of season
quarter
C
Note: Bioclimatic variables were calculated with the “dismo”package in R
(Hijmans et al., 2021), and the last five variables were developed specifically
for winter-annual grass species.
ECOSPHERE 5of22

We checked for sign changes or differences that
exceeded ±0.5. From the 24 variables under initial
consideration, this process resulted in seven predictor
variables: mean diurnal range, isothermality, tempera-
ture seasonality, mean temperature of the warmest
quarter, mean temperature of the coldest quarter, pre-
cipitation seasonality, and cool season precipitation.
Model development and evaluation
We used an ensemble of boosted regression tree (BRT)
models to model ventenata’s climatically suitable
habitat. BRTs have been demonstrated as a flexible
and performative method for species distribution
modeling, allowing for the use of diverse data types,
handling outliers, and incorporating interactions (Elith
et al., 2008; Norberg et al., 2019). Each model in the
ensemble used a unique combination of occurrence
and pseudo-absence data from the 100 replicate
datasets (species data).
To improve model performance and minimize the
potential for overfitting, we performed a cross-validated
complete grid search of plausible hyperparameter values
for 20 of the 100 replicates. From the tuning results, we
chose the hyperparameters with the greatest ratio of per-
formance to model simplicity (bag fraction =0.75, inter-
action depth =5, minimum observations per node =40,
learning rate =0.005, number of trees =1500). We used
the selected hyperparameters to perform model simplifi-
cation where predictors were iteratively dropped from
the model until the average change in deviance exceeded
the original model SE (Elith et al., 2008). As a result of
this process, isothermality was dropped from the model,
leaving six final variables (Appendix S1: Table S2).
We evaluated model performance in terms of
sensitivity, specificity, and the area under the receiver
operating characteristic curve (AUC) with fivefold
cross-validation of each of the 100 replicate datasets. For
each BRT model, we summarize each evaluation statis-
tic by the mean and SD across that BRT model’sfive
cross-validation datasets. We report the ensemble’sper-
formance as the mean of the 100 BRT models’means
and SDs for each statistic.
Climate suitability and relationships
We predicted climatic suitability across the western
United States under current (PRISM) and future climate
conditions (NEX-DCP30). We assessed the area of
suitable habitat by grouping probability values into four
classes. The high suitability class (0:72 ≤p< 1) was
determined from the threshold that maximized the kappa
statistic (Freeman & Moisen, 2008). We then further
divided the probability gradient below this threshold into
three additional classes: unsuitable (0 ≤p<0:05), low
suitability (0:05 ≤p<0:36), and moderate suitability
(0:36 ≤p<0:72). To identify differences between regions,
we examined classes across the west within and outside
the invaded range, and by US EPA Level III ecoregions
(Omernik & Griffith, 2014).
To characterize relationship between ventenata and
current climate, we examined partial dependence plots
from the BRT models. To determine where and how cli-
mate conditions were influencing suitability, we quanti-
fied the effect size of each predictor on each prediction
location with SHapley Additive exPlanations (SHAP
values; Lundberg et al., 2019). As the calculation of
SHAP values is computationally intensive, we calculated
SHAP values for 10% of BRT models for both current and
future climate scenarios. From these raw SHAP values,
importance was determined by ranking the predictors by
their absolute SHAP values.
Assessing uncertainty and reliability
We used prediction variability and climate novelty to
assess how different types of uncertainty may impact
the model prediction reliability. We quantified both
SDM- (current and future) and GCM-derived prediction
variability. Current SDM-derived prediction variability
was calculated as the SD of the BRT 100 models’
predictions. For future SDM-derived variability, the same
process was followed for each GCM and then averaged
across the 30 GCMs. GCM-derived prediction variability
was calculated as the SD of suitability predictions from
the 30 GCMs. To determine whether and where the
ensemble may be extrapolating, we assessed climate nov-
elty and the source of novelty (i.e., the climate predictor
leading to novel conditions) using multivariate environ-
mental suitability surfaces (MESS; Elith et al., 2010).
Finally, to identify locations where future suitability is
more certain, we summed the number of GCMs that
predicted a location as highly suitable (i.e., “suitability
agreement”) and identified ecoregions within the western
United States where predictions agreed that climate
conditions could support ventenata.
All model development, prediction, and assessment
were performed in R version 4.1 (R Core Team, 2020)
using packages caret (Kuhn, 2022), dismo (Hijmans
et al., 2021), gbm (Greenwell et al., 2020), raster
(Hijmans, 2022a), spThin (Aiello-Lammens et al., 2015),
terra (Hijmans, 2022b), and treeshap (Komisarczyk
et al., 2022).
6of22 NIETUPSKI ET AL.

GCM case studies
To evaluate the bounds of potential future climate that
may impact ventenata’s distribution, we chose to high-
light the results from four GCMs (Figure 2). These four
GCMs were selected by examining the change in climate
by the end of the century under RCP8.5 with respect to
the top two predictors (i.e., the two predictors with the
greatest relative importance). RCP8.5 is a high-warming
scenario (Riahi et al., 2011) and may project well to
mid-century (Schwalm et al., 2020, but see Ritchie &
Dowlatabadi, 2017). While less severe warming scenarios
may be more likely (Hausfather & Peters, 2020), the four
selected GCMs project lower changes to mean annual
temperature in comparison with the other GCMs used in
this study.
RESULTS
The BRT models demonstrated strong performance with
an average cross-validated AUC of 0.85 ± 0.02, sensitivity
of 0.95 ± 0.01, and specificity of 0.38 ± 0.06 (individual
replicate evaluation statistics are found in Appendix S1:
Table S3). Climatic suitability was most strongly
explained by cool season precipitation followed closely
by temperature seasonality, which accounted for 55% of
the relative influence (Figure 3A). Mean temperature in
the coldest and warmest quarters accounted for an addi-
tional 31% of relative influence. Precipitation seasonality
and mean diurnal range accounted for the remaining
14% of relative influence. The top four predictors all
showed a roughly unimodal relationship with suitability
(Figure 3B). Suitability decreased with increasing prec-
ipitation seasonality and had a multimodal relationship
with mean diurnal range.
Current suitability and uncertainty
Substantial portions of the western United States were
predicted to contain highly to moderately suitable habitat
(Figure 4;Table2), which occurred between 1000- and
2000-m elevation (interquartile range; average of 1200 m).
Within ventenata’s invaded range (see Figure 1), 25% of
the area was classified as highly or moderately suitable
FIGURE 2 Projected change in temperature seasonality and cool season precipitation from 1982–2012 to 2069–2099 for 30 global
climate models (GCMs). Temperature seasonality and cool season precipitation are the two most important ventenata (Ventenata dubia)
habitat suitability predictors and are not highly correlated with projected mean annual temperature. The lower left quadrant represents
climate most like current conditions and the upper right quadrant represents the greatest divergence in future climate. Gray label shading
and triangles indicate the four GCM case studies.
ECOSPHERE 7of22

habitat. However, these classes cover less than 5% of the rest
of the western United States. Most vulnerable ecoregions
(i.e., with abundant highly and moderately suitable habitat)
overlapped the invaded range. Vulnerable ecoregions out-
side of the invaded range include the Arizona/New Mexico
Mountains (29,505 km
2
; 8% of total), Central Basin and
Range (27,805 km
2
; 8% of total), and the Southern Rockies
(17,095 km
2
; 5% of total). Habitat in these three ecoregions
occurred at an average elevation of 2300 m. SDM replicates
also had low prediction variability across the western
United States (Figure 4;seeAppendixS1:Section4formore
information on SDM-derived uncertainty). Only areas with
moderate suitability, such as portions of the Northern Basin
and Range, had relatively high prediction variability
(e.g., suitability 0.4 ± 0.2).
The importance map illustrates the dominant effect
of cool season precipitation across much of the western
United States (Figure 5). However, suitability in some
regions was more strongly governed by other climate
predictors, some of which were not ranked highly in
the model-level assessment (Figure 3). For example, high
precipitation seasonality (i.e., >80 mm) decreased suit-
ability in much of the Central California Foothills and
Coastal Mountains, Central California Valley, Sierra
Nevada, and Klamath Mountains/California High North
Coast Range ecoregions. In other areas, low coldest
quarter temperature (i.e., <−6C) decreased suitability
along the higher elevation Rockies.
Future suitability and uncertainty
The distribution of climatically suitable habitat varied
substantially by GCM (Figure 6). Projections of highly
suitable habitat may decrease from 19% to 95% by the
end of the century (Appendix S1: Table S7). Moderately
suitable habitat was also frequently projected to decrease
by GCMs, but some GCMs (FIO-ESM, INMCM4.0,
IPSL-CM5B-LR, MIROC5, GFDL-ESM2G, CNRM-CM5,
CESM1-CAM5, and CESM1-BGC) projected an increase
in moderate suitability by as much as 39%. Relative to the
others, these GCMs projected larger increases in cool sea-
son precipitation along with smaller increases in temper-
ature seasonality and smaller changes in mean annual
temperature (Figure 2). Compared with the current
distribution, models driven by some GCMs projected a
similar elevational range but higher average elevation
(e.g., CNRM-CM5: avg. 2200 m and range 1700–2700 m;
MPI-ESM-MR: avg. 1850 m and range 1400–2300 m),
while other GCMs projected a narrower elevational range
FIGURE 3 Partial dependence and relative importance of each climate predictor. (A) The bar width indicates the median relative
importance of the model replicates while the error bars represent the 5th and 95th percentiles. (B) Black lines show the partial dependence
for model replicates and the red line is a loess curve fit to highlight the general trend among replicates. Labels correspond with the numbers
in the bar labels of panel A.
8of22 NIETUPSKI ET AL.

and increased average elevation (Appendix S1:FigureS8).
At higher latitudes of the western United States, highly
and moderately suitable habitats were projected to shift
upward from 250 to 1250 m (GISS-E2-R and CanESM2),
while these classes within the mid-latitudes shifted from
−50 to 1250 m (MPI-ESM-MR and HadGEM2-ES) and
low latitudes shifted from 200 to 850 m (GISS-E2-R and
IPSL-CM5A-LR). Across GCMs, much of the suitable
habitat expansion occurred in the Rockies and especially
the Southern Rockies (e.g., NorESM1-M; Appendix S1:
Figure S6), but some GCMs projected increased suitability
along the entire range (e.g., CNRM-CM5; Appendix S1:
Figure S6).
Future projections also differed significantly between
currently invaded and uninvaded ranges. Within the
invaded range, highly suitable habitat contracted in all
GCMs and moderately suitable habitat contracted in
26 out of the 30 GCMs. Outside of the invaded range,
highly and moderately suitable habitat expanded in about
60% of GCMs and by as much as 160% and 124%,
TABLE 2 Composition of suitability classes within and outside of the invaded range of ventenata (Ventenata dubia) under current
climate conditions (PRISM; 1982–2012).
Climatic
suitability
class
Area inside
invaded range
(km
2
× 1000)
Area outside
invaded range
(km
2
× 1000)
Percentage of area
inside invaded
range
Percentage of area
outside invaded
range
Percentage of area in
western United
States
High 96.97 26.92 9.5 1.3 4.1
Moderate 170.73 60.61 16.7 3.0 7.6
Low 410.13 508.30 40.1 25.2 30.2
Unsuitable 345.17 1419.17 33.7 70.4 58.1
FIGURE 4 Ventenata (Ventenata dubia) climatic suitability and species distribution model-derived prediction variability across the
western United States under current climate (PRISM; 1982–2012). Areas with the white crosshatch pattern indicate climate novelty. SDM,
species distribution model.
ECOSPHERE 9of22

respectively. GCMs that projected an increase in
moderately suitable habitat outside of the invaded range
generally projected an increase across the western
United States. A few GCMs that did not predict an
expansion across the west but did project an expansion
outside the invaded range shared temperature seasonality
and cool season precipitation characteristics with those
that projected an expansion across the west. However,
7 out of 10 of these GCMs (e.g., CanESM2, FGOALS-g2,
HadGEM2-AO, IPSL-CM5A-LR, IPSL-CM5A-MR, MIROC-
ESM, MIROC-ESM-CHEM, NorESM1-M) had greater than
average changes in mean annual temperature (Figure 2).
Variability from GCM projections was nearly an
order of magnitude greater than that derived from the
SDM (Figure 6). GCM-derived variation generally
decreased with decreasing suitability (Table 3), and the
greatest GCM-derived variation occurred along the
Rockies, much of which has low current suitability.
This variation arose from the interaction between cool
season precipitation, temperature seasonality, and mean
annual temperature. In contrast to the GCM-derived
variability, SDM-derived variability was highest in areas
predicted to be highly or moderately suitable under current
climate. This pattern of high SDM-derived uncertainty
FIGURE 5 The source of climate novelty and predictor importance under current climate conditions (PRISM; 1982–2012). Climate
novelty shows the location of climate conditions that were outside the observed range in the occurrence data. Importance shows the predictor
with the greatest influence on the climate suitability. Under current climate conditions, most areas of novel climate were predicted to be
unsuitable (Figure 4). Large contiguous areas of climate novelty in the Northwestern Plains and the Mojave and Sonoran deserts contributed to
the low suitability for ventenata (Ventenata dubia). However, low suitability was most influence by low cool season precipitation in the
Northwestern Plains. Therefore, while temperature seasonality was novel in the Northwestern Plains, the unsuitable classification is mostly
related to cool season precipitation, suggesting that predictions may be reasonable. While the Mojave and Sonoran Deserts were novel in terms
of warmest quarter temperatures, cool season precipitation, and coldest quarter temperatures, the observed ventenata–climate relationship
suggests that habitat suitability is low near the novel values. Therefore, the low suitability predicted in these areas may be reasonable. Not all
locations of climate novelty coincided with low ventenata suitability. Exceptions occurred in areas that were novel due to high diurnal range or
low precipitation seasonality. In these cases, the ventenata–climate relationship shows high suitability at the high or low extremes of the
observed climate gradient (e.g., high suitability at 20C diurnal range). The Arizona/New Mexico Mountains, Arizona/New Mexico Plateau,
and Chihuahuan had diurnal range values (20 and 21C), which increased the predicted suitability. When coupled with the relatively high
species distribution model-derived variability, suitability predictions in these areas may be less reasonable.
10 of 22 NIETUPSKI ET AL.

coincides with future projections of moderate suitability,
indicating that suitability may become marginal in many
currently suitable areas.
Under many of the GCM predictions, cool season
precipitation remained the most important climate pre-
dictor of suitability at the end of the century (Figures 7
and 10C;AppendixS1:FigureS11). Nonetheless, tem-
perature seasonality displaces cool season precipitation
as the most important predictor in much of the
Intermountain West. Additionally, precipitation season-
ality became more important at higher latitudes to the
west of the Sierra Nevada and Cascade ranges, as well as
in some parts of the Mojave, Sonoran, and Chihuahuan
Deserts.
Novel climate conditions were projected to increase
across the west under the majority of the GCMs
(Figure 8; Appendix S1:FigureS5), with some GCMs
(e.g., GFDL-CM3) projecting novel conditions for
much of the western United States. Climate novelty was
projected to mostly expand from the currently novel
regions. For example, novelty from the warmest
quarter temperatures may expand to the north from
the Mojave and Sonoran Deserts. Other notable
expansions occurred for coldest quarter temperatures
along the Coast Range and mean diurnal range in
the Arizona/New Mexico Plateau and Southwestern
Tablelands.
Although ventenata habitat suitability varied across
the study and across GCMs, some locations had high
agreement across multiple GCMs (Figure 9). Agreement
most frequently occurred within the Southern and
Middle Rockies, Wasatch and Uinta Mountains,
and Idaho Batholith ecoregions. When weighted by
ecoregion area (i.e., agreement/unit area), the Canadian
FIGURE 6 Summary of suitability projections from 30 global climate models (GCMs) for the end of the century (2060–2099) based on
the RCP8.5 emissions pathway. GCM suitability (left) is the median suitability across the 30 suitability projections. GCM-derived variability
(center) is the SD of suitability across the 30 suitability projections. Species distribution model-derived variability (right) is the SD of
suitability from boosted regression tree replicates averaged across GCMs. SDM, species distribution model.
TABLE 3 Prediction variability for climatic suitability classes arising from global climate models (GCMs) and the species distribution
model (SDM).
Climatic suitability
class
GCM-derived variability SDM-derived variability
Avg Min Max Avg Min Max
High 0.206 0.038 0.341 0.023 0.011 0.049
Moderate 0.224 0.048 0.343 0.038 0.015 0.070
Low 0.104 0.002 0.338 0.018 0.004 0.067
Unsuitable 0.015 0.000 0.288 0.004 0.001 0.017
Note: Variability was summarized by the median prediction across GCMs for the end of the century (2069–2099) under RCP8.5 climate change scenario.
ECOSPHERE 11 of 22

Rockies, Northern Rockies, North Cascades, Eastern
Cascades Slopes and Foothills, and Blue Mountains
ecoregions had moderate vulnerability to ventenata. The
Arizona/New Mexico Mountains had somewhat low vul-
nerability, with GCM model agreement primarily occur-
ring only in the highest elevations. Similarly, high
elevations of the Central Basin and Range had somewhat
high agreement, but these areas compose a relatively
small proportion of the total ecoregion.
GCM case studies
The four selected GCMs illustrate individual model
output across a range of climate conditions (Figure 10;
Appendix S1: Figure S10). FIO-ESM simulates a future
climate that is the most like present-day conditions and
shows little change in suitability resulting from mild
increases in temperature seasonality and cool season pre-
cipitation. Increases in temperature seasonality under
FIGURE 7 The number of global climate models (GCMs) that ranked each predictor as most important under RCP8.5 at the end of the
century (2069–2099).
12 of 22 NIETUPSKI ET AL.

MPI-ESM-MR resulted in a large suitability decrease both
inside and outside of the invaded range. As expected,
the other two GCMs showed more moderate changes in
suitability within the invaded range and variable
changes outside. At higher latitudes, increased cool
season precipitation increased the proportion of suitable
habitat across the elevation gradient (e.g., CNRM-CM5)
and increased temperature seasonality tended to
decrease the proportion of suitable habitat across the
elevation gradient (e.g., MRI-CGCM3). In the most
extreme example (MPI-ESM-LR), very little suitable
habitat remained at mid- and high latitudes by the end
of the century. However, at lower latitudes, variation in
the change of temperature seasonality and cool season
precipitation between GCMs did not impact the propor-
tion of moderately and highly suitable habitat per eleva-
tion class, which either remained similar or increased in
the future (i.e., bottom row of Figure 10).
FIGURE 8 The number of global climate models (GCMs) with novel future climate conditions under RCP8.5 for the end of the century
(2069–2099).
ECOSPHERE 13 of 22

DISCUSSION
Climate change will influence the distribution and
composition of many plant communities and favor some
non-native species over natives. We found that projected
future climate may lead to substantial changes in the
distribution of ventenata’s climatic habitat under a
high-warming climate scenario (RCP8.5; Riahi et al., 2011),
with suitable habitat shifting toward the mountainous
regions of the west, especially toward the Rocky
Mountains. However, predictions of future habitat varied
greatly among GCMs, highlighting the uncertainty in
future invasion potential. While projections of future habi-
tat were variable, some areas within the west were
projectedtobesuitablebyasmanyas25–30 of the GCMs.
Uncertainty in current and future invasion potential can
be compounded by novel climate conditions, which may
mean that predictions in these areas are unreliable
(Elith et al., 2010). However, based on the observed
ventenata–climate relationship, we found that most novel
climate conditions were far beyond the threshold at which
climate conditions became unsuitable, indicating that pre-
dictions in these areas may be reasonable.
Under both the current and future climate scenarios,
areas with higher cool season precipitation supported
suitable ventenata habitat. Cool season moisture coin-
cides with the growth cycle of ventenata and is critically
important for winter-annual grass species (Bradley, 2009;
Pilliod et al., 2017; Prevéy & Seastedt, 2015). Within the
native range of many winter-annual grasses, the balance
of annual and perennial species abundance follows a
precipitation seasonality gradient, where annuals are
favored by cool season precipitation and higher seasonality
(Clary, 2008). Our results coincide with previous research
in this regard with one exception: locations in California
that have extreme precipitation seasonality have decreased
habitat suitability. This observed effect of precipitation
seasonality in California may be a proxy for competitive
exclusion from other non-native annual grass species
because California’s Mediterranean climate is suitable for
FIGURE 9 Agreement between global climate models (GCMs) of future climatically suitable habitat for ventenata (Ventenata dubia)in
the western United States. Suitability agreement is the count of GCMs for which the location was predicted to be highly suitable at the end
of the century (2069–2099). US EPA level III ecoregions in gray shading indicate greater vulnerability to invasion as determined by the area
weighted sum of suitability agreement. Numeric codes correspond with the ecoregion names listed in the table in Figure 1. Ecoregions with
low vulnerability include 1, 3, 4, 13, 20, and 22; 15 and 18 are moderate; 33 is high and 31 is very high.
14 of 22 NIETUPSKI ET AL.

and heavily invaded by numerous annual grasses
(HilleRisLambers et al., 2010; Jackson, 1985).
While cool season precipitation was the most
important predictor under current climate, future suit-
ability was driven as much, if not more, by temperature
seasonality. We found that high temperature seasonality
decreased habitat suitability, a finding that has also been
reported for other non-native annual grasses in North
America (Bowen & Stevens, 2020). Ventenata’s life cycle
tends to start (Noone et al., 2013) and end later (Pavek
et al., 2011) than other co-occurring invasive annual
grasses, which may expose ventenata to more extreme
cold or heat spells during important phases of its growth
and reproduction. Additionally, if there is a rapid transi-
tion from extreme cold to hot temperatures, the duration
in which soil temperature and moisture requirements
meet ventenata’s needs may not last long enough for
ventenata to complete a life cycle or produce viable seeds.
Future increases in temperature seasonality have also
been projected to drive the future grass–shrub dominance
in dryland ecosystems where shrubs are expected to
be favored over perennial grasses (Gremer et al., 2018).
Therefore, while the abundance and distribution of pre-
cipitation over the year will be important moving for-
ward, the variation in temperature over the year may
have a greater impact on the habitat suitability for certain
species and functional groups.
Unsurprisingly, current predictions indicate that the
largest area of high and moderate suitability is found
within the invaded range. And while the percentage of
suitable habitat in the remaining western United States is
relatively small, the total area of highly and moderately
suitable habitat is nearly equivalent. Given adequate
time, transport, and climate conditions, this could mean
a doubling of the total invaded area within the western
United States. Invasive plant species commonly have a
lag phase of 20 or more years after their introduction
before they rapidly spread to new locations (Aikio
et al., 2010). While ventenata has been present in the
Pacific Northwest for several decades, expansion has only
recently increased in this region (Wallace et al., 2015).
Another invasive annual grass, cheatgrass, rapidly
expanded to its invaded range over about 40 years after
being present in the United States for several decades
prior (Novak & Mack, 2001). Therefore, ventenata may
only be at the beginning stages of a rapid expansion
across the west.
The spatial patterns of current habitat generally
aligned with those of another recently produced SDM
for ventenata (Jarnevich et al., 2021). However, our
FIGURE 10 The distribution of climatic suitability in the western United States by elevation and latitude under current (PRISM;
1982–2012) and future (2069–2099) climate conditions, based on four selected global climate models. (A) The proportion of each elevation
class occupied by each suitability class, and (B) the area of each elevation class. The rows represent three latitude classes (29.3–38.2,
38.2–43.6, and 43.6–49.1, respectively). FIO-ESM has future climate the most similar to current climate. CNRM-CM5 has future climate
the most favorable to ventenata (Ventenata dubia) expansion, driven primarily by increasing cool season precipitation. MRI-CGCM3 is an
intermediate case featuring increasing cool season precipitation and increasing temperature seasonality. MPI-ESM-MR has future climate
the least favorable for ventenata expansion, characterized by increasing temperature seasonality.
ECOSPHERE 15 of 22

models predicted substantial suitable habitat in the
Arizona/New Mexico Mountains but did not predict
much suitable habitat in the Wyoming Basin and
Northwestern Great Plains. Predictions from the Jarnevich
et al. (2021) study contrast with our predictions in these
ecoregions. Unlike Jarnevich et al. (2021), we included
cool season precipitation and temperature seasonality in
our model. Cool season precipitation in the Arizona/New
Mexico Mountains increased suitability while temperature
seasonality decreased suitability in the Wyoming Basin
and Northwestern Great Plains. Nevertheless, our habitat
predictions did overlap with those of Jarnevich et al.
(2021) in parts of the Wyoming Basin and Northwestern
Great Plains and other nearby ecoregions.
Employing a combination of existing and recently
developed methods allowed for a clearer understanding
of uncertainty and the locations vulnerable to current
and future invasion. Model projections based on the
30 GCMs generally agreed that suitability within many
currently suitable areas would decrease, similar to results
reported for other invasive grasses (Bradley, 2009;
Gallagher et al., 2013). Many currently suitable areas are
predicted to transition from high or moderate suitability
to moderate or low suitability. In addition, future
SDM-derived variability is relatively high in many of the
same locations, indicating that habitat may become
marginal in the future (Gould et al., 2014).
Across the western United States, all future suitability
projections resulted in a decrease in the total amount
of highly suitable habitat (Appendix S1: Table S6).
However, suitability agreement suggested that ecoregions
including the Southern Rockies, Middle Rockies,
Wasatch and Uinta Mountains, and Idaho Batholith
would contain substantial amounts of highly suitable
habitat (Figure 9). These same ecoregions and much of
the Rocky Mountains also had the greatest GCM-derived
uncertainty (Figure 6). In these regions, GCMs consis-
tently projected that there would be an increase in
coldest quarter temperatures, which increased the suit-
ability for ventenata. This increase in winter tempera-
tures could lead to more precipitation falling as rain and
a lengthening of the growing season, which has been
shown to increase the habitat suitability for other inva-
sive annual grass species (Concilio et al., 2013). Within
the Middle Rockies and Idaho Batholith ecoregions,
many of the GCMs also projected large increases in
temperature seasonality, which somewhat dampened the
impact of more favorable coldest quarter temperatures.
However, for the Southern Rockies and Wasatch and
Uinta Mountains ecoregions, GCMs projected that
temperature seasonality would not increase as much,
which led to a more frequent prediction of highly suitable
habitat. The Southern Rockies and Wasatch and Uinta
Mountains ecoregions also have considerable habitat
under current climate, so climate may be expected to
remain suitable now and in the coming decades.
Our results support the close examination and
interpretation of model predictions through the lens of
climate novelty. Projections of habitat suitability across
space and time can lead to model transferability issues
arising from altered species–environment relationships in
prediction domains and extrapolation of models in novel
climatic space (Mainali et al., 2015; Qiao et al., 2019). In
our study, climate novelty did not commonly reduce the
reliability of model predictions. This contrasts with some
reported assumptions about model extrapolation outside
the observed environmental range (Gallagher et al., 2013;
Owens et al., 2013). Owens et al. (2013) found that
relationships fit by several modeling algorithms produced
unrealistic responses that suggested high probability
at the margins of the observed environmental range.
As a result, unrealistic predictions occurred when these
models were applied outside the observed environmental
range.
The relationship between ventenata and most of
the climate predictors was nearly unimodal (Figure 3),
suggesting that the SDM may have reasonably captured
the upper and lower suitability thresholds for ventenata.
Consequently, many of the novel climate locations across
the western United States fell outside the observed range
of suitable conditions. For example, warmest quarter
temperatures in the Mojave and Sonoran Deserts were
novel but also far outside the observed climate relation-
ship. However, not all climate variables had a unimodal
relationship. Two of our predictors had relationships
where probability was high at the margins of the
observed environmental range, resulting in extrapolation
when climate conditions were novel. One example of this
extrapolation occurred from the influence of diurnal
range in the Arizona/New Mexico Mountains, Arizona/
New Mexico Plateau, and Chihuahuan Deserts, where
predictions should be cautiously interpreted. In addition
to examining the fitted species–environment relationship,
it was also useful to determine the contribution of
individual climate predictors to suitability using a
recently developed method based on SHAP values. SHAP
values in the Mojave and Sonoran Deserts indicated
that the influence of cool season precipitation and tem-
perature seasonality was four times the influence of the
warmest quarter temperatures (Appendix S1: Figure S12).
Therefore, climate novelty from warmest quarter temper-
atures in this region did not substantially contribute to
the resulting suitability and may not signal unreliable
predictions.
Much of the current moderate and highly suitable
habitat outside the invaded range occurs at higher
16 of 22 NIETUPSKI ET AL.

elevations. This result is unsurprising given that the
invaded range falls within the top half of the western
United States’latitudinal gradient. Ventenata already
invades higher elevations within its invaded range than
other problematic annual grass species, including dry
pine and mixed-conifer forests (Tortorelli et al., 2020). At
lower latitudes, we found that suitable habitat was fre-
quently projected in areas where dense conifer forests
(i.e., areas with high canopy cover) would likely prevent
ventenata establishment or confine populations to forest
openings. Therefore, while the proportion of suitable
habitat in ecoregions like the Arizona/New Mexico
Mountains, Central Basin and Range, and the Southern
Rockies may be relatively high, the actual area where this
species could invade within these ecoregions may be
much smaller.
Future suitable habitat generally increased in eleva-
tion within a given latitudinal band, consistent with
reports for many plant species (Lenoir et al., 2008).
However, in contrast to the lowest latitudinal band where
the proportion of each elevation class was maintained or
increased, the mid- and high latitudinal bands saw
decreases in the area and proportion of elevation classes
with suitable habitat with increasing temperature season-
ality (Figure 10). This pattern suggests that if temperature
seasonality increases, there may be a southward latitudi-
nal shift of moderately or highly suitable habitat.
However, suitability reductions in the northern latitudes
may be tempered if cool season precipitation increases
along with temperature seasonality. Historically, many
species have been projected to shift both up in elevation
and northward in latitude (Hickling et al., 2006;
Pauchard et al., 2016). Our models partially confirm this
effect, where increasing coldest quarter temperatures
increased suitability at northern latitudes and decreased
suitability at southern latitudes. However, as with other
results from this analysis, increasing temperature season-
ality trumped this effect, causing decreased suitability at
northern latitudes.
We acknowledge several caveats in our modeling
approach. While we used the finest resolution gridded
datasets (800 m) available for both current and future
conditions, the resolution may be problematic in complex
or high-relief terrain. We also note that there are other
nonclimate-related barriers to plant invasion that should
be considered, particularly at local scales. Metrics related
to soils and topography were evaluated in the early phases
of our model development but had negligible influence
and were excluded from further consideration. However,
soils and topography may be important for invasion suc-
cess at finer scales. Biotic interactions, such as competition
and propagule pressure, also influence invasion dynamics.
Mature native perennial grasses, especially those with
similar growth forms and phenology, can be highly
effective competitors with some invasive annual grasses
such as cheatgrass (Chambers et al., 2007), although one
study suggests ventenata does not strongly interact with
neighboring herbaceous species (Tortorelli et al., 2022b).
While biotic relationships may be important to consider
at local scales, macroclimate imposes the most control
on species distributions at broad scale, and species
niche relations also appear conserved at broad scales
(Fraterrigo et al., 2014;Sextonetal.,2009;Sober

on &
Nakamura, 2009). Therefore, our models should be
used as an initial coarse filter related to climate habitat
only, with consideration of other processes at finer
scales.
A key caveat of forests’resistance to ventenata
invasion is that higher elevation forests of the western
United States have experienced increasing insect out-
breaks (Jenkins et al., 2014;Saabetal.,2014) and wildfires
(Alizadeh et al., 2021) but are not regenerating (Pettit
et al., 2019; Rodman, Veblen, Battaglia, et al., 2020;
Rodman, Veblen, Chapman, et al., 2020), increasing
vulnerability to invasion. In some cases, western wildfires
are also producing larger, contiguous high-severity patches
(Stephens et al., 2014). Stevens and Latimer (2015)found
that climate conditions, disturbance, and propagule
pressure were the main factors limiting plant invasion
within higher elevation forests. Coldest quarter temp-
eratures are the main limiting factor for ventenata suit-
ability along much of the Rocky Mountains, and most
GCMs projected that coldest quarter temperatures would
decrease in the future, increasing suitability. When com-
bined with expected future disturbance conditions, the
only limitation to the spread of ventenata in these areas
may be propagule pressure.
CONCLUSIONS
The invasion of ventenata in recent decades, its potential
to degrade previously invasion-resistant plant communi-
ties, and the potential effect of grass invasion on fire
behavior suggest that management of this species in
the invaded range may be a priority both now and into
the future. Our models suggest that while the climatic
suitability of the invaded range will likely decrease in
the future, suitable conditions for ventenata will persist.
However, the trend in decreasing habitat suitability in
much of the currently suitable range may provide oppor-
tunities for increased management efficacy (Gallagher
et al., 2013). Areas of high future suitability will tend to
move toward the east, throughout the Rockies, Wasatch,
and Unita Mountains, and up in elevation. Given the pre-
sent trend toward more fire in higher elevation forests
ECOSPHERE 17 of 22

and the reduction in canopy cover resulting from insect
outbreaks, these mountain ranges may be highly suscep-
tible to future ventenata invasion.
ACKNOWLEDGMENTS
This research was supported by an appointment to
the United States Forest Service Research Participation
Program administered by the Oak Ridge Institute for
Science and Education (ORISE) through an interagency
agreement (contract number DE-SC0014664) between
the U.S. Department of Energy (DOE) and the U.S.
Department of Agriculture (USDA). All opinions
expressed in this paper are the authors and do not neces-
sarily reflect the policies and views of the USDA, DOE, or
ORISE. The authors like to thank Michelle Day and
Bridgett Naylor for their assistance with preparation of
the ventenata occurrence data. The authors would
also like to thank Lila Leatherman and Jeff Hollenbeck
for their work on early versions of this analysis.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
DATA AVAILABILITY STATEMENT
Species data and novel code (Nietupski et al., 2024)
are available from Figshare: https://doi.org/10.6084/m9.
figshare.24615732.v2. Climate data (Thrasher et al., 2013)
are available from https://registry.opendata.aws/nasanex/.
ORCID
John B. Kim https://orcid.org/0000-0002-3720-7916
Claire M. Tortorelli https://orcid.org/0000-0001-9493-
9817
Becky K. Kerns https://orcid.org/0000-0003-4613-2191
REFERENCES
Abatzoglou, J. T., and C. A. Kolden. 2011. “Climate Change in
Western US Deserts: Potential for Increased Wildfire and
Invasive Annual Grasses.”Rangeland Ecology & Management
64: 471–78. https://doi.org/10.2111/REM-D-09-00151.1.
Abatzoglou, J. T., and A. P. Williams. 2016. “Impact of
Anthropogenic Climate Change on Wildfire across Western
US Forests.”Proceedings of the National Academy of Sciences of
the United States of America 113: 11770–75. https://doi.org/10.
1073/pnas.1607171113.
Aiello-Lammens, M. E., R. A. Boria, A. Radosavljevic, B. Vilela, and
R. P. Anderson. 2015. “spThin: An R Package for Spatial
Thinning of Species Occurrence Records for Use in Ecological
Niche Models.”Ecography 38: 541–45.
Aikio, S., R. P. Duncan, and P. E. Hulme. 2010. “Lag-Phases in
Alien Plant Invasions: Separating the Facts from the
Artefacts.”Oikos 119: 370–78. https://doi.org/10.1111/j.1600-
0706.2009.17963.x.
Alizadeh,M.R.,J.T.Abatzoglou,C.H.Luce,J.F.Adamowski,
A. Farid, and M. Sadegh. 2021. “Warming Enabled Upslope
Advance in Western US Forest Fires.”Proceedings of the
National Academy of Sciences of the United States of
America 118: e2009717118. https://doi.org/10.1073/pnas.
2009717118.
Applestein, C., and M. J. Germino. 2021. “Patterns of Post-Fire
Invasion of Semiarid Shrub-Steppe Reveals a Diversity of
Invasion Niches within an Exotic Annual Grass Community.”
Biological Invasions 24: 741–759. https://doi.org/10.1007/
s10530-021-02669-3.
Barbet-Massin, M., F. Jiguet, C. H. Albert, and W. Thuiller. 2012.
“Selecting Pseudo-Absences for Species Distribution Models:
How, Where and How Many?”Methods in Ecology and
Evolution 3: 327–338. https://doi.org/10.1111/j.2041-210X.
2011.00172.x.
Bartley, T. J., K. S. McCann, C. Bieg, K. Cazelles, M. Granados,
M. M. Guzzo, A. S. MacDougall, T. D. Tunney, and
B. C. McMeans. 2019. “Food Web Rewiring in a Changing
World.”Nature Ecology & Evolution 3: 345–354. https://doi.
org/10.1038/s41559-018-0772-3.
Bellard, C., B. Leroy, W. Thuiller, J. F. Rysman, and F. Courchamp.
2016. “Major Drivers of Invasion Risks throughout the World.”
Ecosphere 7: 1–14. https://doi.org/10.1002/ecs2.1241.
Boria, R. A., L. E. Olson, S. M. Goodman, and R. P. Anderson. 2014.
“Spatial Filtering to Reduce Sampling Bias Can Improve the
Performance of Ecological Niche Models.”Ecological
Modelling 275: 73–77. https://doi.org/10.1016/j.ecolmodel.
2013.12.012.
Bowen, A. K. M., and M. H. H. Stevens. 2020. “Temperature,
Topography, Soil Characteristics, and NDVI Drive
Habitat Preferences of a Shade-Tolerant Invasive Grass.”
Ecology and Evolution 10: 10785–97. https://doi.org/10.1002/
ece3.6735.
Bradley, B. A. 2009. “Regional Analysis of the Impacts of Climate
Change on Cheatgrass Invasion Shows Potential Risk and
Opportunity.”Global Change Biology 15: 196–208. https://doi.
org/10.1111/j.1365-2486.2008.01709.x.
Bradshaw, C. J. A., B. Leroy, C. Bellard, D. Roiz, C. Albert,
A. Fournier, M. Barbet-Massin, J. M. Salles, F. Simard, and F.
Courchamp. 2016. “Massive Yet Grossly Underestimated
Global Costs of Invasive Insects.”Nature Communications
7(1): 12986. https://doi.org/10.1038/ncomms12986.
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. https://doi.org/10.1641/
0006-3568(2004)054[0677:eoiapo]2.0.co;2.
Chambers, J. C., B. A. Roundy, R. B. Blank, S. E. Meyer, and
A. Whittaker. 2007. “What Makes Great Basin Sagebrush
Ecosystems Invasible by Bromus tectorum?”Ecological
Monographs 77: 117–145. https://doi.org/10.1890/05-1991.
Clary, J. 2008. “Rainfall Seasonality Determines Annual/Perennial
Grass Balance in Vegetation of Mediterranean Iberian.”Plant
Ecology 195: 13–20.
Concilio, A. L., M. E. Loik, and J. Belnap. 2013. “Global Change
Effects on Bromus tectorum L. (Poaceae) at Its High-Elevation
Range Margin.”Global Change Biology 19: 161–172. https://
doi.org/10.1111/gcb.12032.
Creutzburg, M. K., A. C. Olsen, M. A. Anthony, J. D. Maestas,
J. B. Cupples, N. R. Vora, and B. W. Allred. 2022. “A Geographic
Strategy for Cross-Jurisdictional, Proactive Management of
Invasive Annual Grasses in Oregon.”Rangelands 1–8: 173–180.
https://doi.org/10.1016/j.rala.2021.12.007.
18 of 22 NIETUPSKI ET AL.

Crystal-Ornelas, R., E. J. Hudgins, R. N. Cuthbert, P. J. Haubrock,
J. Fantle-Lepczyk, E. Angulo, A. M. Kramer, et al. 2021.
“Economic Costs of Biological Invasions within North
America.”NeoBiota 67: 485–510. https://doi.org/10.3897/
neobiota.67.58038.
D’Antonio, C., and P. Vitousek. 1992. “Biological Invasions by
Exotic Grasses, the Grass/Fire Cycle, and Global Change.”
Annual Review of Ecology and Systematics 23: 63–87. https://
doi.org/10.1146/annurev.es.23.110192.000431.
Dewitz, J. 2019. “National Land Cover Database (NLCD) 2016
Products: U.S. Geological Survey Data Release.”https://doi.
org/10.5066/P96HHBIE.
Elith, J., M. Kearney, and S. Phillips. 2010. “The Art of Modelling
Range-Shifting Species.”Methods in Ecology and Evolution 1:
330–342. https://doi.org/10.1111/j.2041-210x.2010.00036.x.
Elith, J., J. R. Leathwick, and T. Hastie. 2008. “A Working Guide to
Boosted Regression Trees.”The Journal of Animal Ecology 77:
802–813. https://doi.org/10.1111/j.1365-2656.2008.01390.x.
Feng, X., D. S. Park, C. Walker, A. T. Peterson, C. Merow, and
M. Pape¸s. 2019. “A Checklist for Maximizing Reproducibility
of Ecological Niche Models.”Nature Ecology & Evolution 3:
1382–95. https://doi.org/10.1038/s41559-019-0972-5.
Fordham, D. A., C. Bertelsmeier, B. W. Brook, R. Early, D. Neto,
S. C. Brown, S. Ollier, et al. 2018. “How Complex Should
Models Be? Comparing Correlative and Mechanistic Range
Dynamics Models.”Global Change Biology 24: 1357–70.
https://doi.org/10.1111/gcb.13935.
Fraterrigo, J. M., S. Wagner, and R. J. Warren. 2014. “Local-Scale
Biotic Interactions Embedded in Macroscale Climate Drivers
Suggest Eltonian Noise Hypothesis Distribution Patterns for
an Invasive Grass.”Ecology Letters 17: 1447–54.
Freeman, E. A., and G. G. Moisen. 2008. “A Comparison of the
Performance of Threshold Criteria for Binary Classification in
Terms of Predicted Prevalence and Kappa.”Ecological
Modelling 217: 48–58. https://doi.org/10.1016/j.ecolmodel.
2008.05.015.
Funk, J. L., I. M. Parker, V. Matzek, S. L. Flory, E. T. Aschehoug,
C. M. D’Antonio, W. Dawson, D. M. Thomson, and J. Valliere.
2020. “Keys to Enhancing the Value of Invasion Ecology
Research for Management.”Biological Invasions 22: 2431–45.
https://doi.org/10.1007/s10530-020-02267-9.
Gallagher, R. V., D. Englert Duursma, J. O’Donnell, P. D. Wilson,
P. O. Downey, L. Hughes, and M. R. Leishman. 2013. “The
Grass May Not Always Be Greener: Projected Reductions in
Climatic Suitability for Exotic Grasses under Future Climates
in Australia.”Biological Invasions 15: 961–975. https://doi.org/
10.1007/s10530-012-0342-6.
Gallien, L., T. Münkemüller, C. H. Albert, I. Boulangeat, and W.
Thuiller. 2010. “Predicting Potential Distributions of Invasive
Species: Where to Go from Here?”Diversity and Distributions
16: 331–342. https://doi.org/10.1111/j.1472-4642.2010.
00652.x.
Gardner, A. S., I. M. D. Maclean, and K. J. Gaston. 2019. “Climatic
Predictors of Species Distributions Neglect Biophysiologically
Meaningful Variables.”Diversity and Distributions 25: 1318–33.
https://doi.org/10.1111/ddi.12939.
Gaskin, J. F., E. Espeland, C. D. Johnson, D. L. Larson,
J. M. Mangold, R. A. McGee, C. Milner, et al. 2021. “Managing
Invasive Plants on Great Plains Grasslands: A Discussion of
Current Challenges.”Rangeland Ecology & Management 78:
235–249. https://doi.org/10.1016/j.rama.2020.04.003.
Gorelick, N., M. Hancher, M. Dixon, S. Ilyushchenko, D. Thau, and
R. Moore. 2017. “Google Earth Engine: Planetary-Scale
Geospatial Analysis for Everyone.”Remote Sensing of
Environment 202: 18–27. https://doi.org/10.1016/j.rse.2017.
06.031.
Gould, S. F., N. J. Beeton, R. M. B. Harris, M. F. Hutchinson,
A. M. Lechner, L. L. Porfirio, and B. G. Mackey. 2014. “A Tool
for Simulating and Communicating Uncertainty when
Modelling Species Distributions under Future Climates.”Ecology
and Evolution 4: 4798–4811. https://doi.org/10.1002/ece3.1319.
Greenwell, B., B. Boehmke, J. Cunningham, and G. Developers.
2020. “gbm: Generalized Boosted Regression Models.”https://
cran.r-project.org/web/packages/gbm/.
Gremer, J. R., C. Andrews, J. R. Norris, L. P. Thomas,
S. M. Munson, M. C. Duniway, and J. B. Bradford. 2018.
“Increasing Temperature Seasonality May Overwhelm Shifts
in Soil Moisture to Favor Shrub over Grass Dominance in
Colorado Plateau Drylands.”Oecologia 188: 1195–1207.
https://doi.org/10.1007/s00442-018-4282-4.
Guillera-Arroita, G., J. J. Lahoz-Monfort, J. Elith, A. Gordon,
H. Kujala, P. E. Lentini, M. A. Mccarthy, R. Tingley, and
B. A. Wintle. 2015. “Is my Species Distribution Model Fit for
Purpose? Matching Data and Models to Applications.”Global
Ecology and Biogeography 24: 276–292. https://doi.org/10.1111/
geb.12268.
Guisan, A., and W. Thuiller. 2005. “Predicting Species Distribution:
Offering More than Simple Habitat Models.”Ecology Letters 8:
993–1009. https://doi.org/10.1111/j.1461-0248.2005.00792.x.
Hausfather, Z., and G. P. Peters. 2020. “Emissions –The “Business
as Usual”Story Is Misleading.”Nature 577: 618–620. https://
doi.org/10.1038/d41586-020-00177-3.
Hickling, R., D. B. Roy, J. K. Hill, R. Fox, and C. D. Thomas. 2006.
“The Distributions of a Wide Range of Taxonomic Groups Are
Expanding Polewards.”Global Change Biology 12: 450–55.
https://doi.org/10.1111/j.1365-2486.2006.01116.x.
Higgins, R. W., Y. Yao, and X. L. Wang. 1997. “Influence of
the North American Monsoon System on the U.S. Summer
Precipitation Regime.”Journal of Climate 10: 2600–2622.
https://doi.org/10.1175/1520-0442(1997)010<2600:
IOTNAM>2.0.CO;2.
Hijmans, R. 2022a. “raster: Geographic Data Analysis and
Modeling.”https://cran.r-project.org/web/packages/raster/.
Hijmans, R. 2022b. “terra: Spatial Data Analysis.”https://cran.r-
project.org/web/packages/terra/.
Hijmans, R. J., S. Phillips, J. Leathwick, and J. Elith. 2021. “Package
‘dismo.’” https://cran.r-project.org/web/packages/dismo/.
HilleRisLambers, J., S. G. Yelenik, B. P. Colman, and J. M. Levine.
2010. “California Annual Grass Invaders: The Drivers or
Passengers of Change?”Journal of Ecology 98: 1147–56.
https://doi.org/10.1111/j.1365-2745.2010.01706.x.
Hulme, P. E. 2017. “Climate Change and Biological Invasions:
Evidence, Expectations, and Response Options.”Biological
Reviews 92: 1297–1313. https://doi.org/10.1111/brv.12282.
Jackson, L. E. 1985. “Ecological Origins of California’s
Mediterranean Grasses.”Journal of Biogeography 12: 349–361.
Jarnevich, C. S., H. R. Sofaer, and P. Engelstad. 2021. “Modelling
Presence Versus Abundance for Invasive Species Risk
ECOSPHERE 19 of 22

Assessment.”Diversity and Distributions 27: 2454–64. https://
doi.org/10.1111/ddi.13414.
Jenkins, M. J., E. G. Hebertson, and A. Steven Munson. 2014.
“Spruce Beetle Biology, Ecology and Management in the
Rocky Mountains: An Addendum to Spruce Beetle in
the Rockies.”Forests 5: 21–71. https://doi.org/10.3390/
f5010021.
Jones, L. C., N. Norton, and T. S. Prather. 2018. “Indicators of
Ventenata (Ventenata dubia) Invasion in Sagebrush Steppe
Rangelands.”Invasive Plant Science and Management 11: 1–9.
https://doi.org/10.1017/inp.2018.7.
Komisarczyk, K., P. Kozminski, S. Maksymiuk, and P. Biecek. 2022.
“treeshap: Fast SHAP Values Computation for Tree Ensemble
Models.”https://cran.r-project.org/web/packages/treeshap/.
Kramer-Schadt, S., J. Niedballa, J. D. Pilgrim, B. Schröder,
J. Lindenborn, V. Reinfelder, M. Stillfried, et al. 2013. “The
Importance of Correcting for Sampling Bias in MaxEnt Species
Distribution Models.”Diversity and Distributions 19: 1366–79.
https://doi.org/10.1111/ddi.12096.
Kuhn, M. 2022. “caret: Classification and Regression Training.”
https://cran.r-project.org/web/packages/caret/.
Lenoir, J., J. C. Gégout, P. A. Marquet, P. De Ruffray, and H. Brisse.
2008. “A Significant Upward Shift in Plant Species Optimum
Elevation during the 20th Century.”Science 320: 1768–71.
https://doi.org/10.1126/science.1156831.
Linderholm, H. W. 2006. “Growing Season Changes in the Last
Century.”Agricultural and Forest Meteorology 137: 1–14.
Liu, Y., J. Stanturf, and S. Goodrick. 2010. “Trends in Global
Wildfire Potential in a Changing Climate.”Forest Ecology and
Management 259: 685–697. https://doi.org/10.1016/j.foreco.
2009.09.002.
Lundberg, S. M., G. G. Erion, and S. Lee. 2019. “Consistent
Individualized Feature Attribution for Tree Ensembles.”arXiv:
1802.03888.
Lurgi, M., B. C. L
opez, and J. M. Montoya. 2012. “Novel Communities
from Climate Change.”Philosophical Transactions of the Royal
Society B: Biological Sciences 367: 2913–22. https://doi.org/10.
1098/rstb.2012.0238.
Mahony, C. R., A. J. Cannon, T. Wang, and S. N. Aitken. 2017.
“A Closer Look at Novel Climates: New Methods and Insights
at Continental to Landscape Scales.”Global Change Biology
23: 3934–55. https://doi.org/10.1111/gcb.13645.
Mainali, K. P., D. L. Warren, K. Dhileepan, A. McConnachie,
L. Strathie, G. Hassan, D. Karki, B. B. Shrestha, and
C. Parmesan. 2015. “Projecting Future Expansion of Invasive
Species: Comparing and Improving Methodologies for Species
Distribution Modeling.”Global Change Biology 21: 4464–80.
https://doi.org/10.1111/gcb.13038.
Nietupski, T. C., J. Kim, C. Tortorelli, R. Lemons, and B. Kerns.
2024. “An Exploration of Current and Future Niche-Based
Invasion Projections and Uncertainty for an Invasive Grass in
the Western US.”Dataset and Code. Figshare. https://doi.org/
10.6084/m9.figshare.24615732.v2.
Nix, H. A. 1986. “A Biogeographic Analysis of Australian Elapid
Snakes.”In Atlas of Elapid Snakes of Australia, edited by
R. Longmore, 4–15. Canberra: Australian Government
Publishing Service.
Noone, M. D., J. S. Kagan, and E. Nielsen. 2013. Western Juniper
(Juniperus occidentalis) and Invasive Annual Grass Mapping in
Eastern Oregon. Portland, OR: Institute for Natural Resources,
Portland State University 57 pp.
Norberg, A., N. Abrego, F. G. Blanchet, F. R. Adler, B. J. Anderson,
J. Anttila, M. B. Araújo, et al. 2019. “AComprehensive
Evaluation of Predictive Performance of 33 Species Distribution
Models at Species and Community Levels.”Ecological
Monographs 89: 1–24. https://doi.org/10.1002/ecm.1370.
Novak, S. J., and R. N. Mack. 2001. “Tracing Plant Introduction
and Spread: Genetic Evidence from Bromus tectorum
(Cheatgrass).”BioScience 51: 114–122. https://doi.org/10.1641/
0006-3568(2001)051[0114:TPIASG]2.0.CO;2.
Omernik, J. M., and G. E. Griffith. 2014. “Ecoregions of the
Conterminous United States: Evolution of a Hierarchical
Spatial Framework.”Environmental Management 54: 1249–66.
https://doi.org/10.1007/s00267-014-0364-1.
Owens, H. L., L. P. Campbell, L. L. Dornak, E. E. Saupe, N. Barve,
J. Sober
on, K. Ingenloff, et al. 2013. “Constraints on
Interpretation of Ecological Niche Models by Limited
Environmental Ranges on Calibration Areas.”Ecological
Modelling 263: 10–18. https://doi.org/10.1016/j.ecolmodel.
2013.04.011.
Pauchard, A., A. Milbau, A. Albihn, J. Alexander, T. Burgess,
C. Daehler, G. Englund, et al. 2016. “Non-native and Native
Organisms Moving into High Elevation and High Latitude
Ecosystems in an Era of Climate Change: New Challenges for
Ecology and Conservation.”Biological Invasions 18: 345–353.
https://doi.org/10.1007/s10530-015-1025-x.
Pavek, P., J. Wallace, and T. Prather. 2011. Ventenata Biology and
Distribution in the Pacific Northwest. Las Cruces, NM: Western
Society of Weed Science 107 pp.
Peterson, A. T. 2003. “Predicting the Geography of Species’
Invasions Via Ecological Niche Modeling.”Quarterly Review of
Biology 78: 419–433. https://doi.org/10.1086/378926.
Pettit, J. M., J. I. Burton, R. J. DeRose, J. N. Long, and S. L. Voelker.
2019. “Epidemic Spruce Beetle Outbreak Changes Drivers of
Engelmann Spruce Regeneration.”Ecosphere 10: e02912.
https://doi.org/10.1002/ecs2.2912.
Phillips, S. J., M. Dudík, J. Elith, C. H. Graham, A. Lehmann, J.
Leathwick, and S. Ferrier. 2009. “Sample Selection Bias and
Presence-Only Distribution Models: Implications for
Background and Pseudo-Absence Data.”Ecological Applications
19: 181–197. https://doi.org/10.1890/07-2153.1.
Pilliod, D. S., J. L. Welty, and R. S. Arkle. 2017. “Refining the
Cheatgrass–Fire Cycle in the Great Basin: Precipitation
Timing and Fine Fuel Composition Predict Wildfire Trends.”
Ecology and Evolution 7: 8126–51. https://doi.org/10.1002/
ece3.3414.
Prather, T. S., and V. Steele. 2009. “Ventenata Control Strategies
Found for Forage Producers.”
Prevéy, J. S., and T. R. Seastedt. 2015. “Effects of Precipitation
Change and Neighboring Plants on Population Dynamics of
Bromus tectorum.”Oecologia 179: 765–775. https://doi.org/10.
1007/s00442-015-3398-z.
PRISM. 2012. “PRISM Climate Group [www Document].”Oregon
State University. http://prism.oregonstate.edu.
Qiao, H., X. Feng, L. E. Escobar, A. T. Peterson, J. Sober
on, G. Zhu,
and M. Pape¸s. 2019. “An Evaluation of Transferability of
Ecological Niche Models.”Ecography 42: 521–534. https://doi.
org/10.1111/ecog.03986.
20 of 22 NIETUPSKI ET AL.

R Core Team. 2020. R: A Language and Environment for Statistical
Computing. Vienna: R Foundation for Statistical Computing.
Riahi, K., S. Rao, V. Krey, C. Cho, V. Chirkov, G. Fischer,
G. Kindermann, N. Nakicenovic, and P. Rafaj. 2011.
“RCP8.5 –A Scenario of Comparatively High Greenhouse Gas
Emissions.”Climatic Change 109: 33–57. https://doi.org/10.
1007/s10584-011-0149-y.
Ritchie, J., and H. Dowlatabadi. 2017. “Why Do Climate Change
Scenarios Return to Coal?”Energy 140: 1276–91. https://doi.
org/10.1016/j.energy.2017.08.083.
Rodman, K. C., T. T. Veblen, M. A. Battaglia, M. E. Chambers,
P. J. Fornwalt, Z. A. Holden, T. E. Kolb, J. R. Ouzts, and
M. T. Rother. 2020. “A Changing Climate Is Snuffing Out
Post-Fire Recovery in Montane Forests.”Global Ecology and
Biogeography 29: 2039–51. https://doi.org/10.1111/geb.13174.
Rodman, K. C., T. T. Veblen, T. B. Chapman, M. T. Rother,
A. P. Wion, and M. D. Redmond. 2020. “Limitations to
Recovery Following Wildfire in Dry Forests of Southern
Colorado and Northern New Mexico, USA.”Ecological
Applications 30: 1–20. https://doi.org/10.1002/eap.2001.
Root, T. L., J. T. Price, K. R. Hall, S. H. Schneider, C. Rosenzweig,
and J. A. Pounds. 2003. “Fingerprints of Global Warming on
Wild Animals and Plants.”Nature 421: 54–57. https://doi.org/
10.1038/nature01309.1.
Saab, V. A., Q. S. Latif, M. M. Rowland, T. N. Johnson,
A. D. Chalfoun, S. W. Buskirk, J. E. Heyward, and
M. A. Dresser. 2014. “Ecological Consequences of Mountain
Pine Beetle Outbreaks for Wildlife in Western North
American Forests.”Forest Science 60: 539–559. https://doi.org/
10.5849/forsci.13-022.
Schwalm, C. R., S. Glendon, and P. B. Duffy. 2020. “RCP8.5 Track
Cumulative CO
2
Emissions.”Proceedings of the National
Academy of Sciences of the United States of America 117:
19656–57. https://doi.org/10.1073/pnas.2007117117.
Seebens, H., T. M. Blackburn, E. E. Dyer, P. Genovesi, P. E. Hulme,
J. M. Jeschke, S. Pagad, et al. 2018. “Global Rise in Emerging
Alien Species Results from Increased Accessibility of New
Source Pools.”Proceedings of the National Academy of Sciences
of the United States of America 115: E2264–E2273. https://doi.
org/10.1073/pnas.1719429115.
Seebens, H., F. Essl, W. Dawson, N. Fuentes, D. Moser, J. Pergl,
P. Pyˇ
sek, et al. 2015. “Global Trade Will Accelerate Plant
Invasions in Emerging Economies under Climate Change.”
Global Change Biology 21: 4128–40. https://doi.org/10.1111/
gcb.13021.
Sexton, J. P., P. J. McIntyre, A. L. Angert, and K. J. Rice. 2009.
“Evolution and Ecology of Species Range Limits.”Annual
Review of Ecology, Evolution, and Systematics 40: 415–436.
https://doi.org/10.1146/annurev.ecolsys.110308.120317.
Shackleton, R. T., B. M. H. Larson, A. Novoa, D. M. Richardson,
and C. A. Kull. 2019. “The Human and Social Dimensions of
Invasion Science and Management.”Journal of Environmental
Management 229: 1–9. https://doi.org/10.1016/j.jenvman.2018.
08.041.
Sober
on, J., and M. Nakamura. 2009. “Niches and Distributional
Areas: Concepts, Methods, and Assumptions.”Proceedings of
the National Academy of Sciences of the United States
of America 106: 19644–50. https://doi.org/10.1073/pnas.
0901637106.
Sofaer, H. R., C. S. Jarnevich, I. S. Pearse, R. L. Smyth, S. Auer,
G. L. Cook, T. C. Edwards, et al. 2019. “Development
and Delivery of Species Distribution Models to Inform
Decision-Making.”BioScience 69: 544–557. https://doi.org/10.
1093/biosci/biz045.
Sorte, C. J. B., I. Ib
añez, D. M. Blumenthal, N. A. Molinari,
L. P. Miller, E. D. Grosholz, J. M. Diez, et al. 2013. “Poised to
Prosper? A Cross-System Comparison of Climate Change
Effects on Native and Non-Native Species Performance.”
Ecology Letters 16: 261–270. https://doi.org/10.1111/ele.12017.
Srivastava, V., V. Lafond, and V. C. Griess. 2019. “Species
Distribution Models (SDM): Applications, Benefits and
Challenges in Invasive Species Management.”CAB Reviews:
Perspectives in Agriculture, Veterinary Science, Nutrition
and Natural Resources 14: 1–13. https://doi.org/10.1079/
PAVSNNR201914020.
Statistics Canada. 2015. “National Road Network.”Statistics
Canada. https://open.canada.ca/data/en/dataset/3d282116-e556-40
0c-9306-ca1a3cada77f.
Stephens, S. L., N. Burrows, A. Buyantuyev, R. W. Gray,
R. E. Keane, R. Kubian, S. Liu, et al. 2014. “Temperate and
Boreal Forest Mega-Fires: Characteristics and Challenges.”
Frontiers in Ecology and the Environment 12: 115–