You are on page 1of 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/358731054

Using environmental niche models to elucidate drivers of the American


bullfrog invasion in California

Article  in  Biological Invasions · June 2022


DOI: 10.1007/s10530-022-02744-3

CITATIONS READS

0 244

2 authors, including:

Nicolette Nelson
University of Connecticut
7 PUBLICATIONS   9 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Nicolette Nelson on 21 February 2022.

The user has requested enhancement of the downloaded file.


Biol Invasions
https://doi.org/10.1007/s10530-022-02744-3

ORIGINAL PAPER

Using environmental niche models to elucidate drivers


of the American bullfrog invasion in California
Nicolette Nelson   · Jonah Piovia‑Scott 

Received: 25 April 2021 / Accepted: 27 January 2022


© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2022

Abstract  Environmental niche models (ENMs) was consistently associated with human activity and
are commonly used to inform management of inva- higher winter temperature. Small pockets of high
sive species, but invasive species often violate  two human activity have sustained bullfrog populations
key  assumptions  of  ENMs which assert that spe- in otherwise unsuitable areas. The areas in California
cies’ ecological niches are stable in space and time, with the highest predicted risk of increased bullfrog
and that organisms are in  equilibrium  with their abundance and future bullfrog expansion include the
environment. The American bullfrog (Rana cates- southern Central Valley, the coast of northern Cali-
beiana) invasion in California provides an excel- fornia, and mid elevations in the mountains of north-
lent opportunity to use ENM approaches recom- ern California. Continued bullfrog invasion in these
mended for overcoming these assumption violations areas is likely to impact sensitive native amphibian
to inform management of a harmful invader. In  the populations.
current study,  we used dynamic ENM approaches
and explicit comparisons between native and invaded Keywords  Environmental equilibrium · Species
niches to: (1) examine how environmental drivers of distribution model · Dynamic environmental niche
occurrence in California shifted during the invasion model · Spatial spread · Human footprint
process, and (2) investigate the potential for contin-
ued bullfrog expansion by explicitly assessing their
environmental equilibrium. Since their initial intro- Introduction
duction to metropolitan areas of California in the late
1800s, bullfrogs have spread throughout the state In order for an invasive species to successfully spread
with the most notable recent expansions in coastal to new areas, the new environment must be acces-
regions, the Central Valley, and the mountains of sible and contain the biotic and abiotic conditions
northern California. Bullfrog occurrence in California required for positive population growth (Soberón and
Peterson 2005; Soberón 2007; Jiménez-Valverde et al.
2011). Invasive species are often able to disperse far-
N. Nelson (*) · J. Piovia‑Scott 
ther and faster than native species, meaning that they
School of Biological Sciences, Washington State
University, Vancouver, WA, USA can access potentially suitable habitats quickly (With
e-mail: nicolette.nelson@uconn.edu 2002; Didham et al. 2007). In addition, invasives tend
to be generalists and are often released from their
N. Nelson 
native predators and competitors, so they are usually
Natural Resources and the Environment, University
of Connecticut, Storrs, CT, USA less limited by biotic interactions than native species

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

(Ehrlich 1989; Arim et  al. 2006; Allen et  al. 2013). all potentially suitable environmental conditions
For these reasons, invasive species can often track (Broennimann and Guisan 2008). However, a global
or adapt to changing environmental conditions bet- description of the niche cannot capture specificities
ter than native species (Ehrlich 1989; Whitney and of a regional invaded niche or explicitly examine the
Gabler 2008). degree to which invasive species are in equilibrium
The relationship between environmental conditions with their environment (Gallien et al. 2012). Explic-
and invasive species spread is typically examined with itly comparing the native and regional invaded
environmental niche models (ENMs), which estimate niches can elucidate the degree to which invasive
a species’ niche from  readily available  data on eco- species are in equilibrium with their environments
logical conditions and  species occurrence (Yalcin and also provide insight into the process of niche
et al. 2017; Thornton and Peers 2019). One potential differentiation during invasion (Gallien et al. 2012).
issue with using these models is that invasive species American bullfrogs  (Rana  catesbeiana, hereaf-
often violate two key assumptions of ENMs: (1) spe- ter “bullfrogs”) are considered one of the world’s
cies’ ecological niches are stable in space and time most harmful invasive species due to their rapid
(Gallien et  al. 2012), i.e., they occupy habitats with global expansion and impacts  on  native species
similar environmental conditions in the native and (Lowe et  al. 2000). They are native to the eastern
invasive portions of their range, and (2) organisms are half of North America but have been introduced to
in  equilibrium  with their environment (Václavík and other parts of the continent and the world through
Meentemeyer 2012), i.e., they have reached all places anthropogenic activities. The bullfrog invasion in
with suitable habitat and are absent from all unsuita- California provides an excellent opportunity to
ble areas (Guisan and Thuiller 2005). Invasive species examine the invasion history and invasion stages of
rarely meet these assumptions because adaptations a harmful species using ENM approaches recom-
acquired  during the invasion process might allow mended for modeling invasive species spread. The
them to occupy new environments (Broennimann invasion has been well documented since the initial
and Guisan 2008; Jiménez-Valverde et  al. 2011) and introduction of bullfrogs to California in the late
because they may not have had enough time to reach 1800s (McKercher and Gregoire 2021), the inva-
all suitable environments (Petitpierre et al. 2012). sive distribution is thought to still be expanding,
Several approaches have been developed to and their native and invaded niches may differ (Fer-
overcome the violation of the environmental equi- nandez and Hamilton 2015; Bissattini and Vignoli
librium assumption when building ENMs for inva- 2017). However,  areas  in the state  that are poten-
sive species (Barbet-Massin et  al. 2018; Srivastava tially suitable to  bullfrogs  but not yet occupied are
et  al. 2019). For example, the predictive power of unknown. Recent studies have described bullfrog
an  ENM  is typically evaluated by determining invasion  history  in the western United States but
how well it predicts the occurrence of points ran- have not used quantitative modeling to elucidate the
domly withheld from the original dataset, but these environmental drivers underlying their expansion
types of metrics (e.g.,  cross-validation) may not (Yap et  al. 2018; McKercher and Gregoire 2021).
be appropriate for invasive species that are not in Bullfrog  management would be aided by updated
equilibrium with their environments (Hattab et  al. information on the patterns underlying their historic
2017). Instead, occurrences from  a  later stage of spread in California, the contribution of different
invasion can be used to assess the predictive power environmental factors to both contemporary and
of ENMs (dynamic environmental niche modeling; historic bullfrog spread, and an examination of their
Ficetola et al. 2010). This dynamic ENM approach environmental equilibrium. In the current study, we
can also be used to examine whether relationships used dynamic ENM approaches and explicit com-
between environmental conditions and invasive spe- parisons between native and invaded niches to: (1)
cies occurrence have shifted during the invasion examine how environmental drivers of occurrence
process. In addition, ENMs for invasive species can in the invaded range shifted during the invasion
be fit with occurrence data from both the native and process, and (2) investigate the potential for contin-
invaded  portions of the range to overcome issues ued bullfrog expansion by explicitly assessing their
with environmental disequilibrium  by capturing environmental equilibrium.

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

Methods gov 2021; McKercher and Gregoire 2021; Table  1).


Additional occurrence data from California were
Background obtained from the California Department of Fish and
Wildlife (Perry 2020). Observations missing geo-
Bullfrogs were initially introduced to California in the graphic coordinates and temporal information were
late 1800s to provide sustenance for California’s rap- removed from the dataset. The positional uncertainty
idly expanding population following the Gold Rush of species occurrence records can substantially reduce
(Jennings and Hayes 1985). Since then, humans have the accuracy of ENMs (Thornton and Peers 2019), so
facilitated their spread by repeatedly transporting occurrences recorded with GPS accuracy greater than
them to new areas and by creating habitats favorable the resolution of environmental data (6 ­km2 for his-
to bullfrog persistence (Peterson et al. 2013). In Cali- toric models and 5 k­ m2 for contemporary models; see
fornia, bullfrogs have expanded from metropolitan below for details) were also removed.
areas to the rest of the state but appear to be absent
from high elevation habitats (McKercher and Gre- Environmental variables
goire 2021). Bullfrogs have contributed to the decline
of several sensitive amphibians in the state (e.g., Important factors to consider when modeling bullfrog
Kupferberg 1997; Yap et al. 2018) and they have the invasion  are  habitat availability, physiological lim-
potential to impact several additional threatened and its, and the timing of juvenile emigration.  Bullfrogs
endangered amphibians if they expand to habitats that are highly aquatic with a multiple-year tadpole stage,
they do not currently occupy (Moyle 1973; Lawler so the availability of permanent waterbodies is likely
et al. 1999; Miaud et al. 2016; Adams 2017). the most important factor contributing to their local
persistence (Peterson et al. 2013; Murray et al. 2015).
Bullfrog occurrence data High mortality has been observed during sudden win-
ter freezes (Willis et al. 1956), but hibernation in deep
Bullfrog  occurrence data (species name  = 
“Litho- permanent waterbodies can buffer these effects (Stew-
bates catesbeianus” or “Rana catesbeiana”) span- art et  al. 2004; Sepulveda and Layhee 2015). Juve-
ning the continental United States were downloaded niles tend to emigrate from breeding ponds  during
from GBIF, BISON, and the USGS Nonindigenous warm rain events in the late summer and fall (Bury
Aquatic Species Database (GBIF 2020; BISON.usgs. and Whelan 1984; Morey 1988). Anthropogenic

Table 1  Variables and data sources used to determine the key drivers of historic  American bullfrog  occurrence  in California,
USA. Some predictor variables were removed from analyses due to high correlations with other predictors
Variable Mechanism of effect on bullfrog occurrence Predicted effect on Data source Resolution1
bullfrog occurrence

Bullfrog occurrence Bullfrog occurrence GBIF


BISON
USGS NAS
CDFW
Elevation Physiological limits (e.g., UV limitation) Negative PRISM 5 ­km2
Minimum winter temperature Physiological limits (winter freezing) Negative PRISM 5 k­ m2
Proximity to waterbodies Habitat availability Positive NHD
Crop cover Historic habitat availability (anthropogenic habitat Positive HYDE 6 k­ m2
modifications)
Human footprint index Recent habitat availability (anthropogenic habitat Positive Dryad 1 ­km2
modifications)
Fall precipitation Juvenile emigration Positive PRISM 5 ­km2
Maximum fall temperature Juvenile emigration Positive PRISM 5 ­km2
1
  Resolution of data before resampling for ENMs

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

activity is likely to increase the rate of bullfrog expan- croplands as well as built environments, human popu-
sion (i.e., Ficetola et  al. 2010) because bullfrogs are lation density, electric infrastructure, pasture lands,
frequently transported by humans, are often actively roads, railways, and navigable waterways. For eleva-
farmed, and because naturalized populations often tion, we used a digital elevation raster downloaded
perform well in human-altered landscapes (Peterson from PRISM (PRISM Climate Group 2020; Table 1).
et al. 2013; Herrel and van der Meijden 2014), espe- Winter freezing was captured by monthly minimum
cially in areas with abundant artificial lentic habitats temperature during the typical bullfrog hibernation
(D’Amore et al. 2009). period (October–March)  downloaded from PRISM
Elevation is another factor that could be linked to (PRISM Climate Group 2020). Climatic conditions
bullfrogs’ physiological limits, habitat availability, affecting juvenile emigration were represented by
and movement. Topographic complexity (i.e., eleva- monthly fall (July–October) precipitation and maxi-
tion change per unit area) and winter freezing may mum temperature data downloaded from PRISM
limit bullfrog colonization of otherwise suitable (PRISM Climate Group 2020).
habitats at high elevations in the Sierra Nevada and
southern Cascades (Peterson et  al. 2013; Sepulveda Environmental niche models
and Layhee 2015). In addition, larval survival may be
impacted by UV-B radiation, which tends to increase Environmental niche models (ENMs) were built to
in intensity with elevation (Garcia et  al. 2015), determine the key environmental drivers of historic
though the hypothesis that UV-B radiation impacts bullfrog spread in California. Individual models were
amphibian survival at high elevations has come under built for discrete historic time periods using hypoth-
scrutiny (e.g., Palen and Schindler 2010). Alterna- esized environmental drivers related to bullfrog habi-
tively, bullfrog habitat suitability may increase with tat availability, physiological limits, and the timing
elevation, especially in the Sierra Nevada, due to the of juvenile emigration. Several metrics were used
prevalence of deep glacial lakes (USGS 2004) and to determine which drivers contributed to bullfrog
snowpack that provides insulation from winter freez- occurrence in each time period.
ing and sustains aquatic habitats during the summer Data  were binned into discrete time periods for
(Rhoades et al. 2016). analysis.  To ensure that ENMs accurately reflected
The availability of bullfrog habitat was represented shifts in historic climate conditions, cutoff points
by geodesic distance to the nearest permanent water- for time periods roughly corresponded to shifts in
body  (lakes, streams, swamps, marshes, reservoirs, climatic regime associated with the Pacific Decadal
and canals, were included; dry lakes, intermittent Oscillation (Newman et al. 2016). Due to low sam-
lakes, and glaciers were excluded) from the National ple size and potential sampling bias in early bull-
Hydrography Dataset (USGS 2004; Table  1); all of frog records, the first two  Pacific Decadal Oscil-
the included habitat types have been associated with lation regimes (1896–1925 and 1926–1945)  were
bullfrogs (Bury and Whelan 1984; Morey 1988; combined, so ENMs were built for the time peri-
Conant and Collins 1991; Stebbins 2003). Man-made ods: 1896–1945, 1946–1980, 1981–2000, and
habitats favorable to bullfrogs (i.e., irrigation ditches 2001–2020.  Climate and cropland data were aver-
and canals that may not have been captured in the aged over the time periods, restricted to the state
USGS dataset) in historic time periods (1896–1945, of California, and then resampled to the spatial
1946–1980, and 1981–2000; see description of time resolution of the variable with the coarsest resolu-
periods below) were represented by the percent cover tion in each dataset (6 ­km2 for historic models and
of cropland in each grid cell (i.e., Ficetola et al. 2010) 5 ­km2 for contemporary models) using bilinear
from the History Database of the Global Environ- interpolation with the projectRaster function in the
ment (Goldewijk et  al. 2011; Table  1). This data- raster R package (Hijmans and van Etten 2012).
set does not represent recent conditions (after 2005) Environmental predictors were also standardized
or non-agricultural man-made habitats. Therefore, for comparability using the normImage function
in the most recent time period (2001–2020), man- in the R package RStoolbox (Leutner et  al. 2019),
made habitats were represented by the Global Human which for each pixel of a raster layer  subtracts the
Footprint Index (Venter et  al. 2016) which includes mean value of the layer and divides by the standard

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

deviation. Prior to building ENMs, bullfrog occur- the models representing the count of non-bullfrog
rences representing each time period were spatially amphibians in each grid cell.
thinned to one occurrence point per grid cell of Models were evaluated  during model building by
environmental data to avoid potential issues with calculating the area under the curve (AUC) and sen-
oversampling (Fourcade et al. 2014). sitivity with five-fold cross-validation. During cross-
Correlations between potential predictor vari- validation, occurrence points were split into five
ables were assessed using Spearman rank cor- roughly equal-sized portions, and then models were
relations for predictor values extracted at bull- fit five times. For each model run, four portions of
frog observations plus 10,000 randomly generated the occurrence points were used as training data and
points for each time period. Relatively high cor- the remaining portion was used as test data (Naimi
relations occurred between elevation and winter and Araujo 2016). Sensitivity requires a threshold to
temperature (r1896-1945 =  − 0.90, r1946-1980 = − 0.89, classify predicted suitability as predicted presence or
r1981-2000 =  − 0.88, r2001-2020 =  − 0.85), elevation absence, which was optimized using the 10th percen-
and fall temperature (r1896-1945 =  − 0.72, r1946-1980 tile threshold, defined as the suitability value below
= − 0.70, r1981-2000 =  − 0.70, r2001-2020 =  − 0.63), and which 10 percent of bullfrog records fell. We chose
winter temperature and fall temperature (r1896-1945 = this threshold because it does not rely on specificity
0.72, r1946-1980 = 0.73, r1981-2000 = 0.74, r2001-2020 = (the probability of correctly classifying absences),
0.66). In order to avoid multicollinearity, we removed which is not suitable for evaluating presence-back-
elevation and fall temperature from our models. ground models where true absences are unknown.
In other words, ENMs were built with the follow- Each models’ ability to predict future invasion was
ing hypothesized environmental drivers of bullfrog evaluated by calculating the AUC, sensitivity, and
spread: crop cover or human footprint, distance to the omission and commission rates. These metrics were
nearest waterbody, minimum winter temperature, and calculated using dynamic environmental niche mod-
mean fall precipitation. eling, where occurrence records from the time period
ENMs were fit for each time period using in question were used as training data and occurrence
the  sdm  R package (Naimi and Araujo 2016), with points from the subsequent time period were used
10,000 pseudo-absence points randomly generated as test data (Runquist et  al. 2019). Because AUC is
from the  study area, the state of California. Models heavily influenced by model parameterization deci-
were run using the generalized linear model (GLM) sions (Lobo et  al. 2008), AUC generated by both
and maximum likelihood methods, with the full evaluation procedures was used to compare perfor-
model structure for each time period (t): mance between models rather than as an absolute

Presence∕pseudo − absencet ∼ 𝛽0 + 𝛽1 (crop covert or human footprintt )


( ) ( )
+ 𝛽2 distance towaterbodiest + 𝛽3 min winter tempt
( ) [ ]
+ 𝛽4 mean fallprecipt + Sampling effort offset + 𝜀t ,

where β represented the relationship between the metric of model performance.


explanatory variables and the presence or absence of The key drivers of historic  bullfrog  occurrence
bullfrogs on the scale of the link function and ε was were determined by examining the AUC and cor-
the binomial error term. Models also included an relation test metrics of variable importance for each
offset term to account for sampling effort; the offset potential environmental driver, extracted from mod-
is a predictor whose coefficient is assumed to be 1 els for each time period using the getVarImp function
(i.e., is not estimated by the model). Sampling effort in the sdm R package (Naimi and Araujo 2016). The
was estimated by downloading all available amphib- AUC method used a cross-validation procedure to
ian records from GBIF (GBIF 2020) for the corre- compare model performance when each variable was
sponding time period, removing bullfrog records, included or excluded (Naimi and Araujo 2016). The
and converting occurrences to a grid with the same correlation test method used a Pearson correlation
spatial resolution as environmental data used to build between standard model predictions (fitted values)

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

and predictions generated from random permutation assess: (1) how well bullfrogs currently fill their pro-
of each variable (Thuiller et  al. 2009). In this case, jected invaded niche in California; (2) how well bull-
low correlation between the predictions indicated frogs currently fill their projected native niche in Cal-
high variable importance. The direction of environ- ifornia; and (3) how (dis-)similar are the invaded and
mental relationships was determined by examining the native niche projections.  Niche filling was deter-
the beta coefficients of each potential environmental mined by  calculating the proportion of bullfrog
driver, extracted from models for each time period occurrences that  fell within areas of predicted pres-
using the sdm R package (Naimi and Araujo 2016). ence from each ENM. The threshold used to bin
predicted suitability from each model into suitable
Assessment of environmental equilibrium or not suitable  for bullfrogs was defined as the ­10th
percentile, the suitability value from each ensemble
In order to assess the environmental equilibrium model below which below which 10 percent of bull-
of  bullfrogs  in California, predictions from the frog records fell. Niche (dis)similarity was assessed
California ENM for the most recent time period by  comparing the predicted suitability from the
(2001–2020) were compared with predictions from native niche model and the California (invaded) niche
an ENM representing the native range niche for the model. Bullfrog  occurrences in California  were then
same time period.  The  native  range ENM was built assigned to the following categories: at equilibrium
from  bullfrog  occurrence records and  environmen- (suitable based on the native and invaded niches),
tal data  representing  the eastern United States,  with colonizing (suitable based on the native niche, unsuit-
the same model specifications and evaluation proce- able based on the invaded niche), adapting (unsuit-
dures described above. The study area extent for the able based on the native niche, suitable based on the
native range model was derived  by adding a buffer invaded niche), or unsuitable (unsuitable based on the
around the native range of bullfrogs (IUCN 2020; native and invaded niches).
Fig.  2) and increasing the buffer size until gaps in Uncertainty in ENM predictions for the contem-
the middle of the eastern US were filled in (110 km porary native and invaded range models was deter-
buffer distance).  100,000 pseudo-absence points mined by calculating the model-based inconsistency
were randomly generated from the study area. The among different model runs for each ENM using
native range ENM was built with the same hypoth- the “entropy” method in the ensemble function in
esized environmental drivers of bullfrog spread as the the sdm R package (Naimi and Araujo 2016), where
California models (correlations between predictors a value of 0 indicated minimum uncertainty (all the
did not exceed 0.7): human footprint, distance to the models predicted the same suitability value) and
nearest waterbody, minimum winter temperature, and a value of 1 indicated maximum uncertainty (half
mean fall precipitation. For the California and native of the models predicted low or high suitability and
range ENMs, ensemble models were created by tak- the other half predicted the opposite value). A mul-
ing the mean of predictions from each of the two tivariate environmental similarity surface (MESS)
modeling methods (GLM and maximum likelihood). was used to determine areas where the native niche
Predicted suitability from the native range ENM was ENM extrapolated outside the range of values in the
then projected onto current (2001–2020) environmen- training dataset, indicating potentially poor predictive
tal conditions in California. power (Elith et al. 2010). The MESS was built using
The native and invaded niches occupied by bull- the mess function in the dismo R package (Hijmans
frogs in California were compared following the et al. 2011).
niche comparison framework outlined by  Gallien
et  al. (2012). Three pairs of data were compared in
geographical space in California:  (1) the observed Results
presences against the California model predic-
tions;  (2) the observed presences against the  native Patterns of historic bullfrog spread
range model predictions; and (3) the invaded against
the  native  model predictions  at the observed pres- We assembled a dataset of 10,284 bullfrog occur-
ences. These comparisons were used  as proxies to rences in California (1,379 after spatial thinning).

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

The number of American bullfrog occurrences metric between different cross-validation runs for
recorded in California has increased roughly expo- each modeling method (Table  2). Under dynamic
nentially since their initial introduction to the state, evaluation, models accurately predicted future bull-
with a large increase in the number of occurrences frog presences with low rates of Type I and Type II
between 2010 and 2020 (Fig.  1). Most of the earli- errors (AUC = 0.78–0.82, sensitivity = 0.90, omis-
est bullfrog occurrences were recorded in the Bay sion = 0.10, commission = 0.0010–0.0042), with little
Area, Central Valley, and within the Los Angeles variation around metrics between different modeling
metropolitan area (Fig. 2). Since then, bullfrogs have methods (Table 3).
expanded in a diffuse pattern to other parts of the Based on the correlation test and area under the
state. The areas with the most notable recent expan- curve (AUC) measures of variable importance, winter
sions are most coastal regions (especially those adja- temperature and human activity were the most con-
cent to metropolitan areas), the Central Valley, and sistent predictors of bullfrog occurrence in California.
the mountains of northern California, including parts Winter temperature was the best predictor of bullfrog
of the Sierra Nevada, Cascades, Klamath, and Coast occurrence in the time periods 1946–1980 (AUC
Ranges (Fig. 2). metric), 1981–2000 (both metrics), and 2001–2020
(AUC metric; Fig. 3), and beta coefficients showed a
Environmental drivers of historic bullfrog spread positive relationship between winter temperature and
bullfrog occurrence (Fig. 3). Croplands were the best
ENMs  built to determine the key drivers of his- predictor of bullfrog occurrence in the time periods
toric bullfrog occurrence  had AUC values rang- 1896–1945 (both metrics) and 1946–1980 (correla-
ing from 0.78—0.86 (Table  2). Based on the ­10th tion test metric; Fig. 3), and bullfrog occurrence was
percentile threshold, models correctly classified positively associated with croplands (Fig.  3). In the
bullfrog presences roughly 90% of the time (sensi- most recent time period (2001–2020), where human
tivity = 0.89–0.90), with little variation around the footprint was used as a metric of human activity

Fig. 1  Cumulative proportion of observations used to examine historic American bullfrog spread in California

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

Fig. 2  Bullfrog occurrences and study areas used to exam- 1946–1980, and 1981–2000; 5 k­ m2 for 2001–2020). The native
ine the spatial patterns underlying historic American bullfrog range study area extent was derived by adding a 110 km buffer
spread in California. Bullfrog occurrences representing each around the native range of bullfrogs, obtained from IUCN
time period were spatially thinned to one occurrence point (IUCN 2020). Bullfrog occurrences in the native range repre-
per grid cell of environmental data (6 k­m2 for 1986–1945, sent the time period 2001–2020

instead of croplands, human footprint contributed waterbody was negatively associated with bullfrog
positively to bullfrog occurrence (correlation test occurrence (Fig. 3).
metric; Fig. 3). Fall precipitation and distance to the
nearest waterbody consistently performed poorly as Assessment of environmental equilibrium
predictors of bullfrog occurrence in California (both
metrics); fall precipitation was positively associated We assembled a dataset of 34,283 bullfrog occur-
with bullfrog occurrence and distance to the nearest rences representing the native range of bullfrogs

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

Table 2  Performance metrics for ENMs used to determine the key drivers of historic American bullfrog occurrence in California,
and in their native range
Time period AUC​ AUC SE Threshold Threshold SE Sensitivity Sensitivity SE

1896–1945 0.85 0.013 0.044 0.013 0.89 0


1946–1980 0.8 0.0084 0.064 0.018 0.90 0
1981–2000 0.78 0.0053 0.13 0.038 0.90 7.2 × ­10–5
2001–2020 0.86 0.004 0.088 0.013 0.90 2.4 × ­10–5
2001–2020 - native 0.79 0.0027 0.051 0.0096 0.90 4.5 × ­10–6
range

Models were evaluated using occurrence data from the time period for which the model was fit. The mean and standard error (SE)
were calculated using values from all cross-validation runs from both modeling methods (GLM and maximum likelihood) for each
time period. AUC​ area under the curve

Table 3  Performance metrics used to determine how well dynamic ENMs predicted American bullfrog occurrence in the following
time period in California
Time period AUC​ Threshold Sensitivity Omission Commission

1896–1945 0.82 0.041 0.90 0.10 0.0010


1946–1980 0.78 0.064 0.90 0.10 0.0042
1981–2000 0.82 0.13 0.90 0.10 0.0041

Models were built using bullfrog occurrences and environmental conditions for the time period listed and were tested using bullfrog
occurrence points for the subsequent time period. The value for each metric represents the mean value from two modeling methods
(GLM and maximum likelihood) for each time period. AUC​ area under the curve

(4,562 points after spatial thinning). The native range Most bullfrog occurrences in California (86% of
ENM representing recent (2000–2020) conditions had observations) occurred in suitable conditions based
an AUC of 0.79; based on the 10th percentile thresh- on both the California and native range ENMs; these
old, the native range model correctly classified bull- observations mainly occurred along the coast of
frog presences 90% of the time (sensitivity = 0.90), California and in the Central Valley, throughout the
with little variation around the metrics between dif- areas predicted as suitable by both models (Fig. 5).
ferent cross-validation runs for each modeling method A small proportion of observations occurred in suit-
(Table 2). Bullfrog occurrence in the native range was able conditions based on the native range ENM but
strongly associated with human footprint (both met- not the California ENM (4% of observations); these
rics; Fig. 3). There was greater uncertainty in predic- were mainly at mid to high elevations in the north-
tions from the native niche model than the California ern Sierra Nevada, Cascades Range, and Klamath
niche model (Fig.  4A). The areas of greatest uncer- Mountains. A small proportion of observations
tainty in the California niche model predictions were occurred in suitable conditions based on the Cali-
mid elevations of mountainous regions (Fig.  4A), fornia ENM but not the native range ENM (4% of
whereas the areas with the greatest uncertainty in the observations); these appeared in small patches in
native niche model predictions were mid to high ele- the southeastern portion of the state and at mid
vations of the Sierra Nevada, Klamath, and Northern elevations in coastal mountains. Observations also
Coast mountain ranges (Fig.  4A). Based on a multi- occurred in areas predicted to be unsuitable by both
variate environmental similarity surface (MESS), California and native range ENMS (6%). These
environmental conditions in California diverged the were dispersed throughout the large portion of the
most from conditions in the native range of bullfrogs state that fell into this category, including high ele-
within the northern Klamath range, with additional vations in the Klamath and coastal ranges, mid to
small areas of divergence scattered throughout the high elevations in the Sierra Nevada and southern
state (Fig. 4B). Cascades, and the Mojave Desert.

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

Fig. 3  Two metrics of variable importance (correlation test for each model (6 ­km2, based on HYDE cropland data for his-
metric and contribution to model AUC) and beta coefficients toric models, and 5 k­ m2, based on PRISM climate variables for
estimated with ENMs of  historic and contemporary  Ameri- contemporary models). Points represent average values from
can bullfrog  occurrence in California. The spatial resolution model runs for both modeling methods (GLM and maximum
of ENMs matched the coarsest available environmental driver likelihood) for each time period

Discussion throughout the state. In recent years they have


moved into coastal and montane areas, while the
Since their initial introduction to urbanized parts of southern deserts and the higher elevation parts of
California in the late 1800s, bullfrogs have spread the southern Sierra Nevada have remained sparsely

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

Fig.  4  A Uncertainty in


predicted suitability from
ENMs describing the
invaded (California-based)
and native niches occupied
by American bullfrogs in
California during the time
period 2001–2020. High
uncertainty (values closer to
1) indicates low agreement
between model runs and
low uncertainty (values
closer to 0) indicates
high agreement between
model runs. B Multivariate
Environmental Similarity
Surface (MESS) comparing
the values of environmental
variables hypothesized to
be important drivers of
America bullfrog spread
between the eastern United
States (native range) and
California (invaded range).
Positive MESS values indi-
cate similarity in conditions
between the two ranges and
negative values indicate
dissimilarity

occupied. Our ENMs suggest that bullfrogs were Habitat associations during the American bullfrog
closely associated with human-modified landscapes invasion of California
and relatively high winter temperature throughout
the invasion. Dynamic ENMs provided reasonable Throughout their invasion of California, bullfrog
predictions for bullfrog suitability over the course occurrence has consistently been associated with
of the invasion, suggesting that this study contrib- higher winter temperature, which primarily occurs
utes to existing literature that shows the utility of at lower elevations in California. Therefore, bull-
dynamic ENMs for characterizing the potential frogs did not appear to benefit from the potential
expansion of invasive species. Explicitly comparing advantages of high elevation habitats including the
bullfrogs’ native and invaded niches in California prevalence of deep glacial lakes (USGS 2004) and
provided reasonable predictions of potential bull- high snowpack that provides insulation from win-
frog spread in the state. ter freezing and sustains aquatic habitats during the

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

Fig.  5  A Habitat suitability of American bullfrogs in Califor- Klamath mountains were predicted as highly suitable based
nia predicted by the invaded (California-based) niche and the on the native range  niche and unsuitable based on the Cali-
native range niche during the time period 2001–2020. B Com- fornia niche (blue). High elevations in the Sierra Nevada and
parison of the native and invaded niches of American bullfrogs southern Cascades were predicted as unsuitable based on both
in California during the time period 2001–2020. Much of the niches (brown), and a small area in northern California was
state was predicted as highly suitable based on the native and predicted to be unsuitable based on the native range niche but
California niches (green). Southern deserts and portions of the highly suitable based on the California niche (tan)

summer (Rhoades et  al. 2016). Instead, bullfrogs from other factors associated with increased eleva-
appeared to be limited in their ability to occupy tion. Studies in other parts of the world invaded by
cold, high elevation habitats, perhaps due to win- bullfrogs found similar associations with elevation
ter freezing at high elevations in the Sierra Nevada (Nori et  al. 2011; Murray et  al. 2015) and winter
and southern Cascades (Peterson et al. 2013; Sepul- temperature (Johovic et  al. 2020), as did a global
veda and Layhee 2015). Since winter temperature study of bullfrog invasions (Ficetola et al. 2007). A
and elevation are highly correlated in California, it thorough examination of the impacts of topographic
is difficult to disentangle the effects of cold winters complexity on regional bullfrog movement (e.g.,

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

Van Buskirk and Jansen van Rensburg 2020) may Insights from dynamic ENMs
help shed light on the role of this factor in limit-
ing bullfrog distributions at higher elevations. The One goal of this study was to use dynamic environ-
fact that winter temperature was not an important mental niche modeling to overcome the violation of
predictor of bullfrog suitability in the native range the ENM assumption of environmental equilibrium.
model is consistent with the idea that bullfrog dis- Overall, California models used to predict future
tribution in California is limited by a unique com- occurrences performed similarly to models trained
bination of temperature, elevation, and topographic on the current dataset. This indicated that dynamic
complexity that may not exist in other portions of ENMs adequately captured shifts in the niches occu-
the bullfrog range. pied by bullfrogs throughout their invasion process
Bullfrogs were associated with humans and and that the dynamic modeling approach was suit-
human-created habitats in California. The consist- able for range-shifting bullfrogs that were not likely
ent association with croplands is consistent with to be in equilibrium with their environments. The
the idea that humans actively introduce bullfrogs in similar performance of these ENMs also means that
populated areas and that bullfrogs are often asso- the California ENM for the most recent time period
ciated with human-made habitat structures (e.g., (2000–2020) may provide insight into areas where
Ficetola et al. 2010; Murray et al. 2015). Moreover, bullfrogs are likely to expand in the future. Future
bullfrog observations in regions of low predicted ENM-based studies of invasive species should incor-
suitability (including the arid eastern regions of the porate dynamic modeling to address the environmen-
state) often had high human footprint values, further tal equilibrium assumption during model evaluation
highlighting the importance of human activity in and to help managers predict future invasions.
sustaining bullfrog populations in otherwise unsuit-
able areas, presumably via repeated introductions Insights from comparing invasive and native range
or the creation of favorable habitats (Peterson et al. ENMs
2013). The relatively high importance of human
footprint in the native range model suggests that the Comparing the native and invaded niches, and the
association between bullfrogs and human-modified frequency of invasive range occurrences in the niche
habitats is not restricted to California; indeed a projections, can also provide insight into the inva-
global analysis also found a strong positive associa- sion process (Gallien et  al. 2012). There was broad
tion with human footprint (Ficetola et al. 2007). overlap between the native and invaded niches sug-
The limited predictive power of waterbodies in gesting that bullfrogs may be approaching environ-
most time periods may indicate that some important mental equilibrium in California. The bulk of bullfrog
bullfrog habitats were not accurately represented by observations in California were in areas designated
this predictor (e.g., cattle ponds or natural peren- as suitable habitat by both the California and native
nial waterbodies used by adults for foraging), espe- range ENMs. These mainly occurred along the coast
cially since those habitats likely occur on a finer and in the Central Valley, areas with a relatively
spatial scale than that captured in our environmental long history of bullfrog occurrences. Most of these
data (Gahl et  al. 2009; Murray et  al. 2015). How- areas were densely occupied by bullfrogs except the
ever, the beta coefficients show a consistent decline southern Central Valley and the coast of northern
in bullfrog habitat suitability with distance from California, suggesting that bullfrogs may continue
water, as expected. The predictor related to the tim- to increase in abundance in these areas in the future.
ing of juvenile emigration, precipitation in the fall Both native and invasive range models generally
(Bury and Whelan 1984; Morey 1988), was gener- predicted low suitability in the arid, eastern parts of
ally a poor predictor of bullfrog occurrence. Taken the state, including the Mojave, Sonoran, and Great
together, the positive associations with cropland, Basin Deserts, as well as parts of the southern Sierra
and the negative associations distance to water may Nevada.
help explain the paucity of bullfrog observations in Areas with high predicted suitability in the native
the arid regions of eastern California. range model but low suitability in the invasive range
model may undergo colonization (Gallien et  al.

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

2012). These areas include large swaths of the moun- Nevada yellow legged frogs (R. sierrae), southern
tains of northern California, including parts of the mountain yellow legged frogs (R. muscosa), and
Coast Ranges, the Klamath Mountains, the southern Yosemite toads (Anaxyrus canorus; Brown et  al.
Cascades Range, and the northern Sierra Nevada 2014, 2015; Pope et  al. 2014). The results of this
(Fig.  5). However, this prediction should be tem- study indicate that bullfrog invasions into the high-
pered by uncertainty associated with extrapolating est elevations of the Sierra Nevada and southern
predicted suitability into areas outside of the ENM Cascades may be slowed or limited by low winter
training dataset (Hartley et  al. 2006; Ficetola et  al. temperatures or other factors associated with high
2007), which occurred when the native range ENM elevations. However, bullfrogs already overlap with
was projected onto environmental conditions in Cali- Cascades frogs at some mid-elevation sites, and our
fornia. Uncertainty from the native range model was results suggest that human activity can contribute to
highest in mountainous regions of northern Califor- the occurrence of bullfrog populations in otherwise
nia (Fig. 4A). The MESS analysis revealed that envi- unsuitable areas. Furthermore, even limited habitat
ronmental conditions in California diverged the most overlap with at-risk native species may facilitate
from conditions in the native range in some of the pathogen transmission from bullfrogs (Peterson
same regions (Fig.  4B). Areas with high predicted and McKenzie 2014; Miaud et  al. 2016; Yap et  al.
suitability in the invasive range model but low suit- 2018). Diseases could then spread to populations
ability in the native range model may suggest niche of native species that do not overlap with bullfrog
shifts during the process of invasion (Gallien et  al. populations. In areas where bullfrogs are known or
2012). These tend to be scattered in relatively small suspected to overlap with sensitive species, careful
patches throughout the state, primarily in arid, low- monitoring will be required to determine the actual
elevation areas. risk that bullfrogs pose to sensitive species on a
local scale. Local-scale monitoring will be espe-
Management implications cially helpful in areas with the highest uncertainty
in model predictions from this study (white areas in
The areas in California with the highest predicted risk Fig. 4A), mountainous regions of northern Califor-
of future bullfrog expansion include low to mid eleva- nia and the eastern Mojave basins.
tion habitats with relatively high winter temperatures Finally, the process of building presence-back-
and human modified habitats, such as the southern ground models such as ENMs includes many addi-
Central Valley, the coast of northern California, and tional sources of uncertainty that should be carefully
mid elevations of the Northern and Southern Coast considered (Beale and Lennon 2012), and many deci-
ranges. High elevation areas of California and the sions were made during the model-building process
Mojave Desert were predicted as unsuitable based that may have impacted the results. Landscape-scale
on both the native and California niches, but bullfrog ENMs such as those developed in this study should
populations have become established in these areas, be paired with regional-scale presence-absence mod-
especially in pockets of suitable habitat with high els, which require repeated systematic sampling but
human activity. Expansion could continue in these may predict invasions for habitat-driven wetland spe-
habitats if bullfrogs adapt to new conditions. cies such as bullfrogs more accurately than ENMs
Competition, predation, and disease transfer (Murray et al. 2015).
from  bullfrogs  have impacted several sensitive
amphibians in California, especially foothill yellow- Acknowledgements  MS committee members Leslie New,
Caren Goldberg, and Daniel Thornton provided guidance, feed-
legged frogs (Rana boylii) and California red-leg- back, and support throughout the duration of this study. Bull-
ged frogs (Rana draytonii; Kupferberg 1997; Dou- frog occurrence records in California were provided by Laura
bledee et al. 2003). Continued bullfrog invasion has Patterson and Heather Perry (California Department of Fish
the potential to impact these species in areas where and Wildlife). Sky Button provided guidance on environmen-
tal niche modeling, and Babak Naimi (University of Helsinki)
bullfrogs are currently rare. Additional threat- provided guidance on use of the sdm R package. The Scien-
ened and endangered amphibians species occur at tific Writing Association for Graduate Students (SWAGS) and
higher elevations in the mountains of California, members of the Piovia-Scott lab at Washington State Univer-
including Cascades frogs (Rana cascadae), Sierra sity gave feedback on writing.

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

Authors’ contributions  Both authors contributed to the outside native ranges and the effect of introduced crayfish.
study conception and design. NN obtained and prepared data, Biol Invasions 19:2633–2646. https://​doi.​org/​10.​1007/​
selected statistical analyses to be used in the study, ran statis- s10530-​017-​1473-6
tical analyses, and produced figures and tables. NN wrote the Broennimann O, Guisan A (2008) Predicting current and future
first draft of the manuscript; JP-S contributed substantially to biological invasions: both native and invaded ranges mat-
revisions of the manuscript. Both authors read and approved ter. Biol Lett 4:585–589. https://​doi.​org/​10.​1098/​rsbl.​
the final manuscript. 2008.​0254
Brown C, Hayes MP, Green G, et al (2015) Yosemite toad con-
Funding  This material is based upon work supported by the servation assessment. USDA Forest Service, CA Depart-
National Science Foundation Graduate Research Fellowship ment of Fish and Wildlife, National Park Service, US Fish
Program under Grant No. 1842493. and Wildlife Service
Brown C, Hayes MP, Green G, Macfarlane D (2014) Mountain
yellow-legged frog conservation assessment for the Sierra
Nevada mountains of California, USA. USDA Forest Ser-
Availability of data and material  This study relies on previ- vice, CA Department of Fish and Wildlife, National Park
ously published datasets that are available from the following Service, US Fish and Wildlife Service
public domain resources: https://​www.​gbif.​org/; https://​bison.​ Bury RB, Whelan JA (1984) Ecology and management of the
usgs.​gov/#​home; https://​nas.​er.​usgs.​gov/​viewer/​omap.​aspx?; bullfrog. US Fish and Wildlife Service, Denver Wildlife
https://​prism.​orego​nstate.​edu/; https://​apps.​natio​nalmap.​gov/​ Research Center
downl​oader/#/; https://​thema​sites.​pbl.​nl/​tridi​on/​en/​thema​sites/​ Conant R, Collins JP (1991) A field guide to reptiles and
hyde/​downl​oad/​index-2.​html; https://​datad​ryad.​org/​stash/​datas​ amphibians: Eastern/Central North America. Houghton
et/​doi:​10.​5061/​dryad.​052q5 Mifflin, Boston
D’Amore A, Hemingway V, Wasson K (2009) Do a threat-
Code availability  The code used for all analyses in this ened native amphibian and its invasive congener differ in
manuscript can be provided by the corresponding author upon response to human alteration of the landscape? Biol Inva-
request. sions 12:145. https://​doi.​org/​10.​1007/​s10530-​009-​9438-z
Didham RK, Tylianakis JM, Gemmell NJ et al (2007) Interac-
Declarations  tive effects of habitat modification and species invasion
on native species decline. Trends Ecol Evol 22:489–496.
Conflict of interest  The authors have no relevant financial or https://​doi.​org/​10.​1016/j.​tree.​2007.​07.​001
non-financial interests to disclose. Doubledee RA, Muller EB, Nisbet RM (2003) Bullfrogs, dis-
turbance regimes, and the persistence of California red-
legged frogs. J Wildl Manag 67:424–438. https://​doi.​org/​
10.​2307/​38027​83
Ehrlich P (1989) Attributes of invaders and the invading pro-
References cesses vertebrates. Biol Invasions Glob Perspect pp
315–328
Elith J, Kearney M, Phillips S (2010) The art of modelling
Adams A (2017) Decline and localized extirpation of the foot-
range-shifting species. Methods Ecol Evol 1:330–342.
hill yellow-legged frog (Rana boylii) in the presence of
https://​doi.​org/​10.​1111/j.​2041-​210X.​2010.​00036.x
the fungal pathogen, Batrachochytrium dendrobatidis:
Fernandez M, Hamilton H (2015) Ecological niche transfer-
Contemporary and historical perspectives. PhD Thesis,
ability using invasive species as a case study. PLoS ONE.
UC Santa Barbara
https://​doi.​org/​10.​1371/​journ​al.​pone.​01198​91
Allen CR, Nemec KT, Wardwell DA et al (2013) Predictors of
Ficetola GF, Maiorano L, Falcucci A et al (2010) Knowing the
regional establishment success and spread of introduced
past to predict the future: land-use change and the distri-
non-indigenous vertebrates. Glob Ecol Biogeogr 22:889–
bution of invasive bullfrogs. Glob Change Biol 16:528–
899. https://​doi.​org/​10.​1111/​geb.​12054
537. https://​doi.​org/​10.​1111/j.​1365-​2486.​2009.​01957.x
Arim M, Abades SR, Neill PE et al (2006) Spread dynamics of
Ficetola GF, Thuiller W, Miaud C (2007) Prediction and
invasive species. Proc Natl Acad Sci 103:374–378. https://​
validation of the potential global distribution of a prob-
doi.​org/​10.​1073/​pnas.​05042​72102
lematic alien invasive species — the American bullfrog.
Barbet-Massin M, Rome Q, Villemant C, Courchamp F
Divers Distrib 13:476–485. https://​doi.​org/​10.​1111/j.​
(2018) Can species distribution models really predict
1472-​4642.​2007.​00377.x
the expansion of invasive species? PLoS ONE. https://​
Fourcade Y, Engler JO, Rödder D, Secondi J (2014) Mapping
doi.​org/​10.​1371/​journ​al.​pone.​01930​85
species distributions with MAXENT using a geographi-
Beale CM, Lennon JJ (2012) Incorporating uncertainty in
cally biased sample of presence data: a performance
predictive species distribution modelling. Philos Trans
assessment of methods for correcting sampling bias.
R Soc B 367:247–258. https://​doi.​org/​10.​1098/​rstb.​
PLoS ONE. https://​doi.​org/​10.​1371/​journ​al.​pone.​00971​
2011.​0178
22
BISON.usgs.gov (2021) BISON occurrence download
Gahl MK, Calhoun AJK, Graves R (2009) Facultative use of
Bissattini AM, Vignoli L (2017) Let’s eat out, there’s cray-
seasonal pools by American bullfrogs (Rana catesbeiana).
fish for dinner: American bullfrog niche shifts inside and
Wetlands 29:697–703. https://​doi.​org/​10.​1672/​08-​56.1

Vol.: (0123456789)
13
N. Nelson, J. Piovia‑Scott

Gallien L, Douzet R, Pratte S et  al (2012) Invasive species Lobo JM, Jiménez-Valverde A, Real R (2008) AUC: a mislead-
distribution models – how violating the equilibrium ing measure of the performance of predictive distribution
assumption can create new insights. Glob Ecol Biogeogr models. Glob Ecol Biogeogr 17:145–151. https://​doi.​org/​
21:1126–1136. https://​doi.​org/​10.​1111/j.​1466-​8238.​2012.​ 10.​1111/j.​1466-​8238.​2007.​00358.x
00768.x Lowe S, Browne M, Boudjelas S, de Poorter M (2000) 100 of
Garcia TS, Rowe JC, Doyle JB (2015) A tad too high: Sen- the world’s worst invasive alien species: a selection from
sitivity to UV-B radiation may limit invasion poten- the Global Invasive Species Database. The Invasive Spe-
tial of American bullfrogs (Lithobates catesbeianus) in cies Specialist Group
the Pacific Northwest invasion range. Aquat Invasions McKercher L, Gregoire DR (2021) Lithobates catesbeianus
10:237–247. https://​doi.​org/​10.​3391/​ai.​2015.​10.2.​12 (Shaw, 1802). In: US Geol. Surv. Nonindigenous Aquat.
GBIF (2020) GBIF occurrence download - Amphibia. Species Database
Accessed 20 Oct 2020 Miaud C, Dejean T, Savard K et  al (2016) Invasive North
Goldewijk KK, Beusen A, de Vos M, van Drecht G (2011) The American bullfrogs transmit lethal fungus Batra-
HYDE 3.1 spatially explicit database of human induced chochytrium dendrobatidis infections to native amphibian
land use change over the past 12,000 years. Glob Ecol host species. Biol Invasions 18:2299–2308. https://​doi.​
Biogeogr 20:73–86 org/​10.​1007/​s10530-​016-​1161-y
Guisan A, Thuiller W (2005) Predicting species distribu- Morey S (1988) American bullfrog (Lithobates catesbeianus).
tion: offering more than simple habitat models. Ecol Lett California Department of Fish and Wildlife California
8:993–1009 Interagency Wildlife Task Group, Sacramento, California
Hartley S, Harris R, Lester PJ (2006) Quantifying uncertainty Moyle PB (1973) Effects of introduced bullfrogs, Rana cates-
in the potential distribution of an invasive species: climate beiana, on the native frogs of the San Joaquin Valley,
and the Argentine ant. Ecol Lett 9:1068–1079. https://​doi.​ California. Copeia 1973:18–22. https://​doi.​org/​10.​2307/​
org/​10.​1111/j.​1461-​0248.​2006.​00954.x 14423​51
Hattab T, Garzon-Lopez CX, Ewald M et al (2017) A unified Murray RG, Popescu VD, Palen WJ, Govindarajulu P (2015)
framework to model the potential and realized distribu- Relative performance of ecological niche and occupancy
tions of invasive species within the invaded range. Divers models for predicting invasions by patchily-distributed
Distrib 23:806–819. https://​doi.​org/​10.​1111/​ddi.​12566 species. Biol Invasions 17:2691–2706. https://​doi.​org/​10.​
Herrel A, van der Meijden A (2014) An analysis of the live 1007/​s10530-​015-​0906-3
reptile and amphibian trade in the USA compared to the Naimi B, Araujo MB (2016) sdm: a reproducible and extensi-
global trade in endangered species. Herpetol J 24:103–110 ble R platform for species distribution modelling. Ecogra-
Hijmans RJ, Phillips AS, Leathwick JR, Elith J (2011) Package phy 39:368–375. https://​doi.​org/​10.​1111/​ecog.​01881
“dismo.” http://​cran.r-​proje​ct.​org/​web/​packa​ges/​dismo/​ Newman M, Alexander MA, Ault TR et al (2016) The pacific
index.​html decadal oscillation, revisited. J Clim 29:4399–4427.
Hijmans RJ, van Etten J (2012) raster: Geographic analysis and https://​doi.​org/​10.​1175/​JCLI-D-​15-​0508.1
modeling with raster data. R Package Version 20–12 Nori J, Akmentins MS, Ghirardi R et al (2011) American bull-
IUCN (2020) Lithobates catesbeianus. In: IUCN Red List frog invasion in Argentina: where should we take urgent
Threat. Species measures? Biodivers Conserv 20:1125–1132. https://​doi.​
Jennings MR, Hayes MP (1985) Pre-1900 overharvest of Cali- org/​10.​1007/​s10531-​011-​0014-3
fornia red-legged frogs (Rana aurora draytonii): The Palen WJ, Schindler DE (2010) Water clarity, maternal behav-
inducement for bullfrog (Rana catesbeiana) introduction. ior, and physiology combine to eliminate UV radiation
Herpetologica 41:94–103 risk to amphibians in a montane landscape. Proc Natl
Jiménez-Valverde A, Peterson AT, Soberón J et al (2011) Use Acad Sci 107:9701–9706. https://​doi.​org/​10.​1073/​pnas.​
of niche models in invasive species risk assessments. 09129​70107
Biol Invasions 13:2785–2797. https://​doi.​org/​10.​1007/​ Perry H (2020) Bullfrog CA locations (compiled from publicly
s10530-​011-​9963-4 available online datasets). California Department of Fish
Johovic I, Gama M, Banha F et al (2020) A potential threat to and Wildlife, West Sacramento, CA
amphibians in the European Natura 2000 network: Fore- Peterson AC, McKenzie VJ (2014) Investigating differences
casting the distribution of the American bullfrog Litho- across host species and scales to explain the distribution
bates catesbeianus. Biol Conserv 245:108551. https://​doi.​ of the amphibian pathogen Batrachochytrium dendroba-
org/​10.​1016/j.​biocon.​2020.​108551 tidis. PLoS ONE. https://​doi.​org/​10.​1371/​journ​al.​pone.​
Kupferberg SJ (1997) Bullfrog (Rana catesbeiana) invasion of 01074​41
a California river: the role of larval competition. Ecology Peterson AC, Richgels KLD, Johnson PTJ, McKenzie VJ
78:1736–1751. https://​doi.​org/​10.​1890/​0012-​9658(1997)​ (2013) Investigating the dispersal routes used by an inva-
078[1736:​BRCIOA]​2.0.​CO;2 sive amphibian, Lithobates catesbeianus, in human-dom-
Lawler SP, Dritz D, Strange T, Holyoak M (1999) Effects of inated landscapes. Biol Invasions 15:2179–2191. https://​
introduced mosquitofish and bullfrogs on the threatened doi.​org/​10.​1007/​s10530-​013-​0442-y
California red-legged frog. Conserv Biol 13:613–622. Petitpierre B, Kueffer C, Broennimann O et al (2012) Climatic
https://​doi.​org/​10.​1046/j.​1523-​1739.​1999.​98075.x niche shifts are rare among terrestrial plant invaders. Sci-
Leutner B, Horning N, Schwalb-Willmann J, Hijmans RJ ence 335:1344–1348. https://​doi.​org/​10.​1126/​scien​ce.​
(2019) Tools for remote sensing data analysis. In: CRAN 12159​33

Vol:. (1234567890)
13
Using environmental niche models to elucidate drivers of the American bullfrog invasion in…

Pope K, Brown C, Hayes M, et al (2014) Cascades frog conser- USGS (2004) National Hydrography Dataset. In: US Dep.
vation assessment. USDA Forest Serivice, Pacific South- Inter. US Geol Surv
west Reserach Station Václavík T, Meentemeyer RK (2012) Equilibrium or not?
PRISM Climate Group (2020). In: Northwest Alliance Com- Modelling potential distribution of invasive species in dif-
put. Sci. Eng. Or. State Univ. http://​prism.​orego​nstate.​edu ferent stages of invasion. Divers Distrib 18:73–83. https://​
Rhoades AM, Huang X, Ullrich PA, Zarzycki CM (2016) doi.​org/​10.​1111/j.​1472-​4642.​2011.​00854.x
Characterizing Sierra Nevada snowpack using variable- Van Buskirk J, Jansen van Rensburg A (2020) Relative impor-
resolution CESM. J Appl Meteorol Climatol 55:173–196. tance of isolation-by-environment and other determinants
https://​doi.​org/​10.​1175/​JAMC-D-​15-​0156.1 of gene flow in an alpine amphibian. Evolution 74:962–
Runquist RDB, Lake T, Tiffin P, Moeller DA (2019) Species 978. https://​doi.​org/​10.​1111/​evo.​13955
distribution models throughout the invasion history of Venter O, Sanderson EW, Magrach A, et al (2016) Data from:
Palmer amaranth predict regions at risk of future inva- Global terrestrial Human Footprint maps for 1993 and
sion and reveal challenges with modeling rapidly shift- 2009, v2, Dryad. Dataset
ing geographic ranges. Sci Rep. https://​doi.​org/​10.​1038/​ Whitney KD, Gabler CA (2008) Rapid evolution in introduced
s41598-​018-​38054-9 species, ‘invasive traits’ and recipient communities: chal-
Sepulveda AJ, Layhee M (2015) Description of fall and winter lenges for predicting invasive potential. Divers Distrib
movements of the introduced American bullfrog (Litho- 14:569–580. https://​doi.​org/​10.​1111/j.​1472-​4642.​2008.​
bates catesbeianus) in a Montana, USA, pond. Herpetol 00473.x
Conserv Biol 10:978–984 Willis YL, Moyle DL, Baskett TS (1956) Emergence, breeding,
Soberón J (2007) Grinnellian and Eltonian niches and geo- hibernation, movements and transformation of the bull-
graphic distributions of species. Ecol Lett 10:1115–1123. frog, Rana catesbeiana, in Missouri. Copeia 1956:30–41.
https://​doi.​org/​10.​1111/j.​1461-​0248.​2007.​01107.x https://​doi.​org/​10.​2307/​14392​41
Soberón J, Peterson AT (2005) Interpretation of models of With KA (2002) The landscape ecology of invasive spread.
fundamental ecological niches and species’ distributional Conserv Biol 16:1192–1203. https://​doi.​org/​10.​1046/j.​
areas 1523-​1739.​2002.​01064.x
Srivastava V, Lafond V, Griess VC (2019) Species distribution Yalcin S, Leroux SJ, Jonathan B (2017) Diversity and suitabil-
models (SDM): applications, benefits and challenges in ity of existing methods and metrics for quantifying species
invasive species management. CAB Rev 14:1–13 range shifts. Glob Ecol Biogeogr 26:609–624. https://​doi.​
Stebbins RC (2003) A field guide to western reptiles and org/​10.​1111/​geb.​12579
amphibians, 3rd edn. Houghton Mifflin Company, Boston Yap TA, Koo MS, Ambrose RF, Vredenburg V (2018) Intro-
Stewart ER, Reese SA, Ultsch GR (2004) The physiology of duced bullfrog facilitates pathogen invasion in the western
hibernation in Canadian leopard frogs (Rana pipiens) United States. PLoS ONE 13:e0188384
and bullfrogs (Rana catesbeiana). Physiol Biochem Zool
77:65–73. https://​doi.​org/​10.​1086/​378921 Publisher’s Note  Springer Nature remains neutral with regard
Thornton DH, Peers MJL (2019) Species distribution mod- to jurisdictional claims in published maps and institutional
eling. In: Population ecology in practice: underused, mis- affiliations.
used and abused methods. Wiley-Blackwell, London
Thuiller W, Lafourcade B, Engler R, Araújo MB (2009) BIO-
MOD – a platform for ensemble forecasting of species
distributions. Ecography 32:369–373. https://​doi.​org/​10.​
1111/j.​1600-​0587.​2008.​05742.x

Vol.: (0123456789)
13

View publication stats

You might also like