ArticlePDF Available

Prediction and validation of the potential global distribution of a problematic alien invasive species - The American bullfrog


Abstract and Figures

Predicting the probability of successful establishment and invasion of alien species at global scale, by matching climatic and land use data, is a priority for the risk assessment. Both large- and local-scale factors contribute to the outcome of invasions, and should be integrated to improve the predictions. At global scale, we used climatic and land use layers to evaluate the habitat suitability for the American bullfrog Rana catesbeiana, a major invasive species that is among the causes of amphibian decline. Environmental models were built by using Maxent, a machine learning method. Then, we integrated global data with information on richness of native communities and hunting pressure collected at the local scale. Global-scale data allowed us to delineate the areas with the highest suitability for this species. Predicted suitability was significantly related to the invasiveness observed for bullfrog populations historically introduced in Europe, but did not explain a large portion of variability in invasion success. The integration of data at the global and local scales greatly improved the performance of models, and explained > 57% of the variance in introduction success: bullfrogs were more invasive in areas with high suitability and low hunting pressure over frogs. Our study identified the climatic factors entailing the risk of invasion by bullfrogs, and stresses the importance of the integration of biotic and abiotic data collected at different spatial scales, to evaluate the areas where monitoring and management efforts need to be focused.
Content may be subject to copyright.
Knowing the past to predict the future: land-use change
and the distribution of invasive bullfrogs
*Dipartimento di Scienze dell’Ambiente e del Territorio, Universita
`di Milano-Bicocca. Piazza della Scienza 1, 20126 Milano,
wLaboratoire d’Ecologie Alpine, UMR CNRS 5553, Universite
´de Savoie, 73376 Le Bourget du Lac cedex, France, zDipartimento di
Biologia Animale e dell’Uomo, Sapienza Universita
`di Roma, viale dell’Universita
`32, 00185 Roma, Italy, §Department of Ecology
and Evolution, University of Lausanne, Biophore Building, CH-1015 Lausanne, Switzerland, }Centre for the Study of
Environmental Change and Sustainability, University of Edinburgh, Drummond Street, EH89XP Edinburgh, UK, kLaboratoire
d’Ecologie Alpine, UMR CNRS 5553, Universite
´Joseph Fourier, Grenoble Cedex 9, France
Biological invasions and land-use changes are two major causes of the global modifica-
tions of biodiversity. Habitat suitability models are the tools of choice to predict potential
distributions of invasive species. Although land-use is a key driver of alien species
invasions, it is often assumed that land-use is constant in time. Here we combine
historical and present day information, to evaluate whether land-use changes could
explain the dynamic of invasion of the American bullfrog Rana catesbeiana (5Litho-
bathes catesbeianus) in Northern Italy, from the 1950s to present-day. We used MAXENT to
build habitat suitability models, on the basis of past (1960s, 1980s) and present-day data
on land-uses and species distribution. For example, we used models built using the 1960s
data to predict distribution in the 1980s, and so on. Furthermore, we used land-use
scenarios to project suitability in the future. Habitat suitability models predicted well the
spread of bullfrogs in the subsequent temporal step. Models considering land-use
changes predicted invasion dynamics better than models assuming constant land-use
over the last 50 years. Scenarios of future land-use suggest that suitability will remain
similar in the next years. Habitat suitability models can help to understand and predict
the dynamics of invasions; however, land-use is not constant in time: land-use modifica-
tions can strongly affect invasions; furthermore, both land management and the suit-
ability of a given land-use class may vary in time. An integration of land-use changes in
studies of biological invasions can help to improve management strategies.
Keywords: alien invasive species, amphibians, future scenarios, habitat suitability models, invasion
dynamics, long term monitoring, Rana catesbeiana, temporal dynamics
Received 17 February 2009 and accepted 5 April 2009
Biological invasions are an unprecedented form of
global change (Ricciardi, 2007), with alien invasive
species (AIS) being a major issue for biodiversity con-
servation at the global scale (Sala et al., 2000; Thuiller,
2007). AIS can negatively impact native species via
predation, competition and diffusion of diseases. More-
over, AIS can strongly affect the environment, for
example by modifying ecosystem functioning and abio-
tic features (Strayer et al., 2006; Ricciardi, 2007). Some
AIS are now present in multiple continents, due to
the interplay between human-assisted diffusion and
natural dispersal ability. This is causing a global
homogenization of faunas and floras, with important
effects on biodiversity pattern at both local scales
and worldwide (McKinney & Lockwood, 1999; Olden
et al., 2004, 2008; McKinney, 2006; Quian & Ricklefs,
Correspondence: Gentile Francesco Ficetola, Dipartimento di
Scienze dell’Ambiente e del Territorio, Universita
`di Milano-
Bicocca. Piazza della Scienza 1, 20126 Milano, Italy, tel. 139 (0)2 64
48 29 45, fax 39 (0)2 64 48 29 96, e-mail:
Global Change Biology (2010) 16, 528–537, doi: 10.1111/j.1365-2486.2009.01957.x
528 r2009 Blackwell Publishing Ltd
The eradication of established AIS can be a difficult
and expensive task (Hulme, 2006). If prevention was
ineffective, the most effective option is often restricting
the spread when the invasions are at their earlier stages.
Much attention is therefore devoted to the understanding
of the dynamic of invasions, to set up plans of biological
screening and prevention in the areas that are most at
risk of invasion (Hulme, 2006). Predictive models are
therefore used to evaluate the areas most at risk of
invasion based on environmental features recorded at
both local and global scale, including climate, land cover
and distribution of other species (Le Maitre et al., 2008).
However, biological invasions are a dynamic process
in which the temporal dimension cannot be overlooked.
Environmental features change in time, species can
quickly evolve in the new environment and these mod-
ifications can influence the invasion dynamics (Urban
et al., 2008). Temporal change of climate is probably the
dynamic feature most often integrated in models of
biological invasions. Climate, in fact, is a major driver
of environmental suitability for AIS, thus many recent
studies used scenarios of future climate to project species
distribution model and to predict the areas that might
become suitable for AIS in the future (Roura-Pascual
et al., 2004; Beaumont et al., 2009; Ficetola et al., 2009).
Climatic suitability is a key tool to predict invasion
risk at coarse spatial scales, i.e., from countrywide to
global (Roura-Pascual et al., 2004; Thuiller et al., 2005).
However, suitable areas identified using bioclimatic
models are often very large, including entire regions
or countries and making it difficult to implement con-
servation plans (Heller & Zavaleta, 2009). Climatic
models should therefore be refined to identify smaller
areas, with the integration of features recorded at finer
scale (Ficetola et al., 2007a).
At finer spatial scales, land-use plays a major role in
the dispersal and distribution of AIS. Land-use is ob-
viously not independent from the distribution of hu-
man population, and AIS often take advantage of
human modified environments (McKinney, 2006).
Land-use is not constant in time, and can change
quickly in response to socioeconomic factors (Falcucci
et al., 2007). Common changes in land-use are the
expansion of urban areas, conversion of natural vegeta-
tion to cropland and pasture or vice versa, and shift of
agricultural practices to increase productivity (Leemans
& Zuidema, 1995; Petit & Lambin, 2002; Hurtt et al.,
2006). Multiple changes in land-use observed at local
scale clearly have a global effect (Leemans & Zuidema,
1995; Hurtt et al., 2006), and the change of land-use is
probably the force most strongly affecting biodiversity
of terrestrial and freshwater ecosystems (Sala et al.,
2000) with important effects on conservation (Maiorano
et al., 2008). Unfortunately, it is often difficult to have
information on past land-use, and predicting future land-
use can be challenging. In practice, most existing studies
using models to predict the distribution of AIS implicitly
assume a constant land-use in the past or in the future
(Domenech et al., 2005 being a noticeable exception).
In this study, we incorporate dynamic land-use in
suitability models, to predict the distribution of a proble-
matic AIS, the American bullfrog Rana catesbeiana Shaw,
1802 ( 5Lithobates catesbeianus) at a regional scale. Bull-
frog is native of eastern North America, but has been
introduced in over 40 countries and four continents
during the 20th century (Lever, 2003). Bullfrog is con-
sidered among the most harmful invasive species, be-
cause it can outcompete and predate native species
(Blaustein & Kiesecker, 2002; Kats & Ferrer, 2003), can
interact with predatory fish which further increase their
negative effect on native amphibians (Blaustein & Kie-
secker, 2002), and it can spread diseases (Garner et al.,
2006). Moreover, only a handful of founders can originate
populations invading large areas in a few generations,
with a spreading ability that challenges the traditional
management plans (Ficetola et al., 2008). Plans are on-
going in several countries to prevent and control the
spread of this species (Lever, 2003; Ficetola et al., 2007b;
Kraus, 2009). Predictive models showed that climatic
features are major drivers of the distribution of invasive
bullfrog populations at coarse spatial scale (Ficetola et al.,
2007a; Giovanelli et al., 2008), but also human activities
and land-use can affect the invasion success of bullfrogs
(Yiming et al., 2006; Ficetola et al., 2007a).
The bullfrog invasion in Northern Italy is documen-
ted since the 1930s (Lanza, 1962; Albertini & Lanza,
1987). This represents a unique opportunity to evaluate
the relationship between land-use changes and biologi-
cal invasions. High quality distribution data cover 70
years of bullfrog expansion; furthermore, land-use in-
formation is available since the 1960s (Falcucci et al.,
2007). This study combined historical information on
land-use and bullfrog distribution, to evaluate whether
land-use can explain bullfrog spread in the investigated
area. The historical information allowed us to test
whether models correctly predict the expansion of this
species. Moreover, we used scenarios of future land-
uses (Rounsevell et al., 2006) to identify the areas that
can be most at risk of invasion in the near future.
Study area and distribution data
We analysed bullfrog distribution in the Eastern River
Po basin, Northern Italy (Fig. 1). The study area is a
lowland dominated by agriculture, with numerous
wetlands, and crossed by a dense network of
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
watercourses and irrigation channels. This is the region
of Europe where bullfrogs are present over the largest
area (Ficetola et al., 2007b). Historical information in-
dicates that a handful of bullfrogs have been first
introduced in a single locality (Fig. 1a) in a single event,
during the 1930s (between 1932 and 1937: Albertini,
1970); genetic analyses confirmed a single introduction
of very few individuals (Ficetola et al., 2008). For the
period 1937–1986, we obtained point data on bullfrog
distribution from the literature (Lanza, 1962; Albertini,
1970, 1983; Albertini & Lanza, 1987). These authors
monitored the spread of bullfrogs using a combination
of field surveys over the whole study area and inter-
views of local people. For the period 1986–2007, we
used an updated version of the database of Ficetola et al
(2007a). We then divided the invasion in three temporal
steps, corresponding to the time frames covered by the
available land-use data: (1) 1937–1965 (98% of data
collected during 1952–1965), hereafter: Distrib1960; (2)
1967–1985 (the majority of data collected during the
1980s), thereafter: Distrib1985; (3) 1987–2007 (95% of
data collected after 1990), thereafter: Distrib2000. Mod-
ifications of the temporal windows by 2–3 years would
produce identical distributions.
Past land-use
We used three land-use/land-cover maps covering the
study area and spanning the time frame 1960–2000. The
oldest map (geographic scale 1 : 200 000; legend of 22
classes) was produced by the National Research Coun-
cil using cadastral datasets collected during the period
1956–1968. We will refer to this map as the ‘Map1960’.
The 1985 and the 2000 land-use maps (hereafter:
Corine1985 and Corine2000) are part of the Corine Land
Cover program started in 1985 by the European
Fig. 1 (a, c, e): Observed distribution of bullfrogs in three temporal steps (1960, 1985 and 2000), and (b, d) suitability predicted using
MAXENT models, based on the distribution observed in the previous temporal step, taking into account land-use change. The star in (a)
represents the locality of first introduction. The different suitability thresholds (0.13 and 0.27) correspond to the 10th percentile training
presence thresholds of models (Pearson et al., 2007).
530 G. F. FICETOLA et al.
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
Community to generate digital land-use/land-cover
maps covering the European continent. These two maps,
produced using satellite images taken during late 1980s
and 1999–2001, respectively (Buttner et al., 2004), have a
legend of 44 classes and a spatial detail comparable to
that of a paper map on a scale of 1: 100 000. The map
obtained using images taken during late 1980s is usually
referred to as Corine1990. However, in this study we
name it Corine1985 for consistency with the available
distribution data. A more detailed description of the
three maps can be obtained from Falcucci et al. (2007). To
obtain a common legend (thematic generalization: Petit
& Lambin, 2002), we reclassified the three maps simpli-
fying the legend already proposed by Falcucci et al.
(2007) and considering five classes: crops (except rice
fields), rice fields, forests, artificial areas, inland water.
Moreover, to obtain three spatially homogeneous layers,
we used a Block Statistics function in ARCGIS 9.2 (ESRI
Redlands, CA, USA), producing three raster maps (3 km
cell size) indicating for each pixel the percentage occu-
pied by each land-use/land-cover class.
Future land-use
Scenarios of future (2020) land-use change were devel-
oped by previous studies (Rounsevell et al., 2006) on the
basis of an interpretation of five alternative storylines of
the ALARM project (Spangenberg, 2007), representing
different assumptions about demographic changes,
technological and socioeconomic development (Nakice-
novic & Swart, 2000). The five scenarios where: GRAS, a
future world orientated towards economic growth ap-
proximating the scenario A1F1 of International Panel on
Climate Change special report on emission scenarios
(SRES); BAMBU ( 5A2 SRES), a continuation into the
future of currently known socioeconomic and policy
strategies; BAMBU-SEL, the same as BAMBU plus
shock in energy price level; BAMBU-CANE, the same
as BAMBU plus contagious natural epidemic; SEDG
(5B1 SRES), a scenario focused on the achievement of
sustainable development. The scenarios were down-
scaled at the spatial resolution of Corine2000 (250 m)
using a mix of spatial multinomial logistic regression
and Bayesian statistics (Dendoncker et al., 2006). Further
details are described elsewhere (Dendoncker et al., 2006;
Rounsevell et al., 2006). For the purpose of this study,
future land-uses were aggregated to 3 km resolution
raster maps, as described for the past land-uses.
Data analysis
We modelled environmental suitability using MAXENT
3.2.1 (Phillips et al., 2006; Phillips & Dudı
´k, 2008).
MAXENT is a machine learning method that estimates
the distribution of a species by finding the probability
distribution of maximum entropy, subject to constraints
representing our incomplete information about the dis-
tribution. The constraints are that the expected value of
each environmental variable should match its average
over sampling locations derived from environmental
layers (Phillips et al., 2006). The model evaluates the
suitability of each grid cell as a function of environ-
mental variables. MAXENT requires presence-only data,
and can calculate the relative importance of different
environmental features (Phillips et al., 2006). We used a
logistic output of MAXENT, with suitability values ran-
ging from 0 (unsuitable) to 1 (optimal habitat) (Phillips
& Dudı
´k, 2008). In recent comparisons, MAXENT was
among the most effectives methods of species distribu-
tion modelling, and showed high quality performance
with both small and large sample sizes (Elith et al., 2006;
Wisz et al., 2008). The reliability of MAXENT has been
confirmed by its capacity to predict the outcome of
introductions of invasive species outside the native
range (Ficetola et al., 2007a) and novel presence local-
ities for poorly known species (Pearson et al., 2007).
We used the following procedure to evaluate the role
of land-use in the bullfrog invasion. First, for each
temporal step, we used MAXENT to build a model relat-
ing species distribution to land-use (per pixel: crop-
land%; rice field%, forest%, artificial areas%, inland
water%) and altitude (average altitude obtained from
a digital terrain model with 20 m cell size). We did not
include climatic features, since coarse scale models
showed that the whole study area has very high cli-
matic suitability (Ficetola et al., 2007a); models includ-
ing scenarios of climate change suggest that climatic
suitability within the study area will remain high in the
future (G. F. Ficetola, unpublished results). We did not
include water bodies distribution because the small
wetlands used by bullfrogs (e.g., ditches, ponds) are
not recorded at the resolution of the historical and
future land-uses; small water bodies used for irrigation
are present at high density in the areas with rice fields
or croplands. We included altitude as it can influence
frog dispersal and the features of water bodies (e.g.,
slow stream ditches are only present in lowlands). We
developed each model using linear, quadratic and hinge
functions (Phillips & Dudı
´k, 2008). Then, we used the
land-use of the subsequent temporal step to predict the
new bullfrog distribution (See Table 1). Therefore, we
built suitability model for Distrib1960 using the
Map1960 (Model1960); then we used Corine1985 to
project suitability in the subsequent temporal step
(Model1960 !85) (Table 1, Fig. 1a and b). Similarly,
we built the model for Distrib1985 using Corine1985
(Model1985); then we used Corine2000 to project
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
suitability (Model1985 !2000) (Table 1, Fig. 1c and d).
Finally, we built the model for Distrib2000 using
Corine2000 (Model2000) and we projected the suitabil-
ity into the future using the five 2020 scenarios (Table 1).
Rice fields where a category not available in the future
scenarios, therefore for Model2000 we pooled rice fields
with other croplands. Pooling rice fields with croplands
did not affect these models, because the models based
on Corine2000, pooling or not pooling rice fields with
croplands, were identical (results not shown). In each
model, we assumed that a cell was suitable for bullfrog
presence if its suitability score was greater than the 10th
percentile of training presence points (Pearson et al.,
2007). We then examined the omission and commission
error of models to evaluate their predictive performance
´nez-Valverde et al., 2008).
We restricted our analysis to an area of high biocli-
matic suitability (180 km 132 km) individuated by
Ficetola et al. (2007a), surrounding the introduction
point. Genetic data showed that bullfrog dispersion
can occur at this spatial scale (Austin et al., 2004). As
there are no major barriers for dispersion (e.g., moun-
tains), we assume that the whole area can be potentially
colonized in a few generations.
We used two methods to evaluate the ability of our
models to predict the bullfrog spread. First, we used a
test (1 df) to compare observed frequencies of correct
and incorrect predictions, and therefore to evaluate if
our models predict distribution in the subsequent tem-
poral step significantly better than expected under
random expectations (Roura-Pascual et al., 2004). Sec-
ond, we built logistic regression models relating the
observed bullfrog distribution in a temporal step to the
suitability predicted on the basis of the two models built
for the previous step (e.g., we predicted Distrib1985 on
the basis of either Model1960 and Model1960 !85). We
assessed significance of the logistic regression using
likelihood ratio. We then used an information-theory
approach, based on Akaike Information Criterion (AIC)
to compare the relative ability of models (Burnham &
Anderson, 2002), and therefore to evaluate if taking into
account land-use changes actually improves model
predictions. This analysis assumes pseudo-absences in
cells where bullfrogs have not been observed. Despite
that pseudo-absences are not always reliable (Engler
et al., 2004), we used this approach to compare the
performance of different models on the same distribu-
tion data, therefore the bias caused by pseudo-absences
was constant between models compared.
During the period 1937–1965, bullfrogs were recorded
in 43 pixels (3 km 3 km) (Fig. 1a); most of presence
localities were clumped close to the introduction local-
ity. During the period 1966–1986, bullfrogs’ distribution
was less clumped and the species was present in 64
pixels. During the period 1987–2007, bullfrog presence
has been recorded in 51 pixels.
Predicting the past invasion
The model built for 1960s showed that bullfrog presence
was associated to high rice fields%, low elevation,
intermediate/high cropland%, and intermediate levels
of forest%; rice field% was the variable with the largest
contribution (56% of explained variation) (Table 2).
Model1960 !85 predicted an expansion of areas sui-
table for bullfrogs, mostly south-east to the area of
introduction. The observed bullfrog expansion followed
the same overall pattern predicted by the model (Fig. 1b
and c).
The model built using Distrib1985 showed that bull-
frogs were associated to intermediate/high cropland%
and low elevation (Table 2). In this model, rice fields
and forest% explained only a minor percentage of
variation (3% or less). Model1985 !2000 predicted an
expansion of areas suitable for bullfrogs, mostly in
the southern and in the northern part of the study area
(Fig. 1d and e). Distrib2000 only slightly expanded the
area of occurrence observed in Distrib1985. Neverthe-
less, it should be noted that the localities with new
Table 1 Distribution data and environmental layers used to build the models predicting bullfrog expansion
Model Input data
Environmental layers
For model calibration For model projection
Model1960 Distrib1960 Map1960 Map1960
Model1960 !1985 Distrib1960 Map1960 Corine1985
Model1985 Distrib1985 Corine1985 Corine1985
Model1985 !2000 Distrib1985 Corine1985 Corine2000
Model2000 Distrib2000 Corine2000 Corine2000
Model2000 !2020 Distrib1985 Corine2000 5 future scenarios
See the text for the acronyms.
532 G. F. FICETOLA et al.
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
records are within the high probability areas predicted
by Model1985 !2000 (Fig. 1d and e).
All models tended to predict well observed presence
localities (Table 2). Conversely, all models tended to
overpredict suitable areas, i.e., predicted suitability in a
large number of pixels where bullfrog presence has not
been recorded yet (Fig. 1, Table 2).
Using logistic regression, Model1960 !1985 pre-
dicted Distrib1985 significantly better than random ex-
pectations (likelihood ratio test, w
547.845, Po0.001).
Its performance was much better than the one of the
model not taking into account temporal change in
environmental variables (difference in AIC,
DAIC 521.9) (Table 3). Similarly, Model1985 !2000
predicted Distrib2000 significantly better than random
expectations (w
520.280, Po0.001). Its performance
was better than that of the model not taking into
account changes in environmental variables, but the
difference was limited (DAIC 50.75) (Table 3). The
small difference between the two models probably
occurred because land-use and consequently suitability
did not change considerably between 1980s and 2000
(Falcucci et al., 2007). For instance, the suitability
predicted by Model1985 was very similar to that pre-
dicted by Model1985 !2000 (r50.97). Conversely,
land-use strongly changed from 1960 to 1985 (as de-
scribed in Falcucci et al., 2007), and the correlation
between Model1960 !1985 and Model1960 was
weaker (r50.54).
Scenarios of future invasion
Model2000 was very similar to Model1985 (Figs 1 and 2,
Table 2), and showed that bullfrogs were associated to
intermediate and high cropland percentage and low
elevation. As the previous models, Model2000
predicted presence with high accuracy, but predicted
suitability in many pixels without bullfrog records
(Table 2).
The projection using the 2020 scenarios did not show
strong changes in suitability. The results were extremely
similar among the five alternative scenarios, with only
minor differences (Fig. 2). These results were essentially
linked to the high stability in land-use projection pre-
dicted for our study area for the timeframe considered,
where intensive agriculture is and will continue to be
the dominant land-use class.
Suitability models, based on land-use and distribution
data, predicted the invasion of bullfrogs in Northern
Italy relatively well. Our analysis used information
collected in three subsequent temporal steps: we pre-
dicted bullfrog distribution in the 1980s using data
collected during the 1960s; similarly, we predicted pre-
sent-day distribution using the 1980s data. This ap-
proach is a true validation of the models; our results
therefore provide a measure of the effectiveness of
Table 2 Predictive performance of models, evaluated by examining omission and commission errors
Model Environmental variables
Omission error Commission error
Model1960 Rice fields, elevation, cropland,
43 39 (11) 4 (32) 1837 (1809) 581 (609) 99.6 o0.001
Model1960 !1985 64 47 (20) 17 (44) 1660 (1633) 737 (764) 52.3 o0.001
Model1985 Elevation, cropland 64 58 (34) 6 (30) 1146 (1122) 1251 (1275) 37.0 o0.001
Model1985 !2000 51 42 (27) 9 (24) 1158 (1142) 1252 (1267) 18.5 o0.001
Model2000 Elevation, cropland 51 45 (26) 6 (25) 1190 (1171) 1220 (1238) 28.3 o0.001
The table reports the correct and incorrect predictions of presence and absence, and Pearson’s w
with 1 df (Roura-Pascual et al.,
2004). Suitability of each cell was based on the 10th percentile of training presence points (Pearson et al., 2007). In parenthesis, the
values under random expectations.
Variables accounting for45% of explained variation. The variables with the largest independent contributions are first in the lists.
Table 3 Relative performance of GLMs, based on MAXENT
models, in predicting bullfrog expansion, evaluated using
Akaike’s Information Criterion (AIC)
to be predicted
change? Model AIC
Distrib1985 Yes Model1960 !1985 549.6
No Model1960 571.5
Distrib2000 Yes Model985 !2000 480.06
No Model1985 480.81
The two models including temporal changes in environmental
variables (Model1960 !1985, Model1985 !2000) are com-
pared against models not including the temporal changes
(Model1960, Model1985).
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
suitability models trying to predict the spread of in-
vasive species. Habitat suitability models are often used
for this task, with applications in risk assessment or in
control strategies (e.g., Ward, 2007; Evangelista et al.,
2008; Nielsen et al., 2008; Ficetola et al., 2009). However,
long term historical data are seldom available (Loo et al.,
2007; Vallecio et al., 2009), and it is therefore difficult
to evaluate whether model predictions correspond to
the actual invasion dynamics. Our analysis suggests
that suitability models have a good performance in
predicting invasions. Nevertheless, the comparison
of predictions with the actual invasion dynamics re-
veals important points. First, the match between pre-
dicted suitability and invasion data was not perfect,
highlighting potential issues. Second, the most impor-
tant variables to explain the bullfrogs’ distribution
were not the same in the three temporal steps, confirm-
ing the complexity of extrapolating model results in
time (Guisan & Thuiller, 2005). Most importantly,
despite altitude does not change in time, other environ-
mental variables (land-use) are not constant in time; the
incorporation of environmental changes in suitability
models can be essential to correctly understand inva-
sion dynamics.
Models and real invasion: reliability and discrepancies
The predictions of our models showed high sensitivity,
i.e., most of presence records corresponded to high
suitability pixels, on the basis of models developed in
the previous temporal step (Table 2). Bullfrogs were
strongly associated to lowland agricultural areas; for
the 1960s model, rice fields were the land-use with the
highest suitability (Table 2). Water bodies are particu-
larly important for bullfrogs: tadpoles usually require 2
years for metamorphosis, and the adults are strongly
dependent on waterbodies (Graves & Anderson, 1987).
In the study area, agriculture is associated to a dense
network of ditches and reservoirs: bullfrogs can take
advantage of the increased availability of permanent
wetlands used for irrigation (Maret et al., 2006). Climatic
suitability models have been deemed to predict suit-
ability at a too coarse spatial scale, with the delinea-
tion of too large areas having limited usefulness for
conservation actions (Heller & Zavaleta, 2009). For
example, based on climatic data, the whole study area
had a high suitability for bullfrogs (Ficetola et al.,
2007a). The integration of climatic data with land-use
or other data recorded at finer scale can help to refine
predictive models, and to focus on the areas where
invasion risk is high at a spatial scale more appropriate
for conservation.
While our models showed high sensitivity, they all
tended to overpredict suitability: bullfrogs have never
Fig. 2 (a) Observed distribution of bullfrog for the temporal
step 2000, and environmental suitability of Model2000. (b–e):
projected suitability in the future (2020) using five scenarios of
land-use change.
534 G. F. FICETOLA et al.
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
been recorded in large areas, which are predicted as
suitable (Table 2, Fig. 1). These commission errors may
have occurred for several reasons. First, the knowledge
of bullfrog distribution is imperfect, because fine scale
monitoring has not been performed over the whole
study area. Efforts to improve our knowledge are cur-
rently ongoing (Ficetola et al., 2007b; Societas Herpetolo-
gica Italica, 2008), but several years will be needed for a
complete knowledge of bullfrog distribution. Moreover,
a model can fail because of the lack of key predictors
affecting distribution, including the presence of preda-
tors/competitors. Our models were limited by the un-
availability of some environmental variables in maps of
past and future land-use. Furthermore, invasive species
can be absent from suitable areas where they have not
been introduced, or where they have not been able to
disperse. In principle, bullfrogs might colonize the whole
study area in a few generations, given their known
dispersal ability (Austin et al., 2004). However, animal
movements are strongly influenced by landscape fea-
tures (Be
´lisle, 2005; Fahrig, 2007). Amphibians are parti-
cularly susceptible to fragmentation and to the presence
of barriers (Cushman, 2006): the study area is strongly
modified by humans, and elements such as roads and
urbanization may have precluded colonization (Ficetola
et al., 2007c).
Finally, the models developed for the three temporal
steps were not identical. Two variables were consistently
important in all models (i.e., elevation and cropland
presence; Table 2). Conversely, rice fields were the most
important variable in the 1960s, while they were unim-
portant in the subsequent temporal steps. The change of
the explanatory power of rice field is probably related to
strong modifications in agricultural practices. In Italy,
rice cultivation traditionally requires the flooding of
fields, and the presence of associated water reservoirs.
Until the 1960s, the permanence of deep water within
rice fields allowed farmers to perform aquaculture with-
in the rice fields; bullfrogs and other amphibians took
advantages of this environment (Albertini, 1970; Lupot-
to, 2005). In the last decades, agricultural practices
strongly changed and the new technologies have mod-
ified the suitability of given land-use classes for many
species. In fact, new rice cultivars require less water,
with fields and irrigation network retaining water for
shorter times and at lower depths, thus strongly redu-
cing the ecological value of rice fields for amphibians
(Lupotto, 2005). As a consequence, the simple presence
of a land-use category (rice field in our case) can be a
misleading indicator of habitat suitability, because key
ecological attributes related to land management have
been changed. This is a further example of the complex-
ities and pitfalls linked to extrapolating model predic-
tions in time.
Integrating temporal variations of the environment
The models including temporal changes of land-use
showed a better prediction of the invasion dynamics
and of the changes in distribution occurring through
time. The global environment is changing at an unprece-
dented rate, and land-use changes have major impacts on
biodiversity distribution (Sala et al., 2000). To date, most
attention has been devoted on the effect of land-use
change on native biodiversity. However, land-use is also
important for the establishment and the spread of AIS.
Land-use modifications have therefore important impli-
cations for the study and management of AIS.
First, land-use modifications can strongly affect
the dynamics of invasive species. For example, in
Model1960 bullfrogs were strongly associated to rice
fields, and the model predicted high suitability in the
East of the study area, where rice field density was high
(Fig. 1b). However, the abundance of rice fields declined
in the study area in the last decades of the 20th century,
and this probably reduced the eastward bullfrog expan-
sion. Both the land cover (i.e., the abundance of rice
fields) and the land management practices (i.e., the way
rice fields are cultivated) are not constant in time, and
are subjected to the constraints of regional planning.
Therefore, an integration of control strategies with the
planning of land-use can have an important role for the
management of invasive species, and to reduce their
spread. Furthermore, ecological models are often used
to predict invasion dynamics, with important applica-
tions in conservation strategies. However, species dis-
tribution models usually assume a static land-use.
When these data are available, taking into account
environmental modification can greatly improve model
performance. Similarly, the combination of models with
scenarios of future environmental changes can provide
important insights that can be used to drive conserva-
tion strategies and regional planning. Our results there-
fore call for an increased consideration of temporal
change of environmental variables when modelling
distribution and suitability.
Our future land-uses do not constitute a ‘prediction’,
but are the outcomes that can arise under different
assumptions, and with a degree of coherence in the
trends of future development (Rounsevell et al., 2006).
Our models showed good performance in predicting
historical changes of bullfrog distribution (Fig. 1), and
our results were stable under various future scenarios
(Fig. 2), suggesting that our conclusion on are robust.
Nevertheless, projection of suitability models outside the
area of calibration are always challenging (Guisan &
Thuiller, 2005), and should be considered with cautions.
For example, the dynamic of the study area can be altered
by unpredicted factors: human-related, climatic or biotic.
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
Despite being human dominated, the study region is
adjacent to some of the areas of Italy with the highest
biodiversity irreplaceability values (Maiorano et al.,
2007), particularly for freshwaters. These unique fresh-
water communities are threatened by the joint effects of
land modifications and invasive species (Gherardi et al.,
2008). As the spread of AIS is strongly related to land-
use changes, an appropriate planning, with coordina-
tion between conservation and development policies,
can help to achieve the management targets, with
optimization of resources.
We thank two anonymous reviewers for comments on an earlier
version of the manuscript. GFF was founded by funded by a
scholarship of the University of Milano-Bicocca, and by a grant of
the French Ministry for Research for young foreign researchers;
WT and LM received support from the EU FP6 ECOCHANGE
integrated project (Challenges in assessing and forecasting biodi-
versity and ecosystem changes in Europe, No: 066866 GOCE) and
CM from EU FPE MARIE CURIE Programme (Biological inva-
sions: patterns and processes. An integrative approach with the
Bullfrog R. catesbeiana in Canada and Europe, OIF No: 022190).
Albertini G (1970) Sulla diffusione della rana toro (Rana catesbei-
ana Shaw) importata nel mantovano. Atti e Memorie della
Accademia di Agricoltura Scienze e Lettere di Verona,20, 67–106.
Albertini G (1983) La Rana catesbeiana Shaw nella bassa pianura
reggiano-modenese. Atti e Memorie della Accademia di Agricol-
tura Scienze e Lettere di Verona,33, 125–154.
Albertini G, Lanza B (1987) Rana catesbeiana Shaw, 1802 in Italy.
Alytes,6, 117–129.
Austin JD, Lougheed SC, Boag PT (2004) Controlling for the
effects of history and nonequilibrium conditions in gene flow
estimates in northern bullfrog (Rana catesbeiana) populations.
Genetics,168, 1491–1506.
Beaumont L, Gallagher R, Thuiller W, Downey P, Leishman M,
Hughes L (2009) Different climatic envelopes among invasive
populations may lead to underestimations of current and
future biological invasions. Diversity and Distributions,15,
´lisle M (2005) Measuring landscape connectivity: the challenge
of behavioral landscape ecology. Ecology,86, 1988–1995.
Blaustein AR, Kiesecker JM (2002) Complexity in conservation:
lessons from the global decline of amphibian populations.
Ecology Letters,5, 597–608.
Burnham KP, Anderson DR (2002) Model Selection and Multimodel
Inference: A Practical Information-Theoretic Approach. Springer
Verlag, New York.
Buttner G, Feranec J, Jaffrain G, Mari L, Maucha G, Soukup T
(2004) The CORINE Land Cover 2000 project. EARSeL
eProceedings,3, 331–346.
Cushman SA (2006) Effects of habitat loss and fragmentation on
amphibians: a review and prospectus. Biological Conservation,
128, 231–240.
Dendoncker N, Bogaert P, Rounsevell M (2006) A statistical
method to downscale aggregated land use data and scenarios.
Journal of Land Use Science,1, 63–82.
Domenech R, Vila M, Pino J, Gesti J (2005) Historical land-use
legacy and Cortaderia selloana invasion in the Mediterranean
region. Global Change Biology,11, 1054–1064.
Elith J, Graham CH, Anderson RP et al. (2006) Novel methods
improve prediction of species’ distributions from occurrence
data. Ecography,29, 129–151.
Engler R, Guisan A, Rechsteiner L (2004) An improved approach
for predicting the distribution of rare and endangered species
from occurrence and pseudo-absence data. Journal of Applied
Ecology,41, 263–274.
Evangelista PH, Kumar S, Stohlgren TJ, Jarnevich CS, Crall AW,
Norman JB, Barnett DT (2008) Modelling invasion for a habitat
generalist and a specialist plant species. Diversity and Distribu-
tions,14, 808–817.
Fahrig L (2007) Non-optimal animal movement in human-
altered landscapes. Functional Ecology,21, 1003–1015.
Falcucci A, Maiorano L, Boitani L (2007) Changes in land-use/
land-cover patterns in Italy and their implications for biodi-
versity conservation. Landscape Ecology,22, 617–631.
Ficetola GF, Bonin A, Miaud C (2008) Population genetics reveals
origin and number of founders in a biological invasion.
Molecular Ecology,17, 773–782.
Ficetola GF, Thuiller W, Miaud C (2007a) Prediction and valida-
tion of the potential global distribution of a problematic alien
invasive species – the American bullfrog. Diversity and Dis-
tributions,13, 476–485.
Ficetola GF, Coı
¨c C, Detaint M, Berroneau M, Lorvelec O, Miaud
C (2007b) Pattern of distribution of the American bullfrog Rana
catesbeiana in Europe. Biological Invasions,9, 767–772.
Ficetola GF, Garner TWJ, De Bernardi F (2007c) Genetic diversity,
but not hatching success, is jointly affected by post glacial
colonization and isolation in the threatened frog, Rana latastei.
Molecular Ecology,16, 1787–1797.
Ficetola GF, Thuiller W, Padoa-Schioppa E (2009) From introduc-
tion to the establishment of alien species: bioclimatic differ-
ences between presence and reproduction localities in the
slider turtle. Diversity and Distributions,15, 108–116.
Garner TWJ, Perkins MW, Govindarajulu P, Seglie D, Walker S,
Cunningham AA, Fisher MC (2006) The emerging amphibian
pathogen Batrachochytrium dendrobatidis globally infects intro-
duced populations of the North American bullfrog, Rana
catesbeiana.Biology Letters,2, 455–459.
Gherardi F, Bertolino S, Bodon M et al. (2008) Animal xenodi-
versity in Italian inland waters: distribution, modes of arrival,
and pathways. Biological Invasions,10, 435–454.
Giovanelli JGR, Haddad CFB, Alexandrino J (2008) Predicting
the potential distribution of the alien invasive American
bullfrog (Lithobates catesbeianus) in Brazil. Biological Invasions,
10, 585–590.
Graves BM, Anderson SH (1987) Habitat suitability index models:
bullfrog, US Fish and Wildlife Service Biological Report 82.
Guisan A, Thuiller W (2005) Predicting species distribution:
offering more than simple habitat models. Ecology Letters,8,
Heller NE, Zavaleta ES (2009) Biodiversity management in the
face of climate change: a review of 22 years of recommenda-
tions. Biological Conservation,142, 14–32.
536 G. F. FICETOLA et al.
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
Hulme PE (2006) Beyond control: wider implications for the
management of biological invasions. Journal of Applied Ecology,
43, 835–847.
Hurtt GC, Frolking S, Fearon MG et al. (2006) The underpinnings
of land-use history: three centuries of global gridded land-use
transitions, wood-harvest activity, and resulting secondary
lands. Global Change Biology,12, 1208–1229.
´nez-Valverde A, Lobo JM, Hortal J (2008) Not as good as
they seem: the importance of concepts in species distribution
modelling. Diversity and Distributions,14, 885–890.
Kats LB, Ferrer RP (2003) Alien predators and amphibian de-
clines: review of two decades of science and the transition to
conservation. Diversity and Distributions,9, 99–110.
Kraus F (2009) Alien Reptiles and Amphibians a Scientific Compen-
dium and Analysis. Springer, Dordrecht.
Lanza B (1962) On the introduction of Rana ridibunda Pallas and
Rana catesbeiana Shaw in Italy. Copeia,1962, 642–643.
Le Maitre DC, Thuiller W, Schonegevel L (2008) Developing an
approach to defining the potential distributions of invasive
plant species: a case study of Hakea species in South Africa.
Global Ecology and Biogeography,17, 569–584.
Leemans R, Zuidema G (1995) Evaluating changes in land cover
and their importance for global change. Trends in Ecology and
Evolution,10, 76–81.
Lever C (2003) Naturalized Amphibians and Reptiles of the World.
Oxford University Press, New York.
Loo SE, Keller RP, Leung B (2007) Freshwater invasions: using
historical data to analyse spread. Diversity and Distributions,13,
Lupotto E (2005) La sperimentazione agraria nella valorizza-
zione del ruolo multifunzionale del sistema risaia. In: Le rane in
risaia: tradizione, scienza e risorsa, (eds Andreone F, di Trana CG,
Lussich E, Tinarelli A, Varala GG), pp. 45–53. Vercelli, Provin-
cia di Vercelli.
Maiorano L, Falcucci A, Boitani L (2008) Size-dependent resis-
tance of protected areas to land-use change. Proceedings of the
Royal Society B-Biological Sciences,275, 1297–1304.
Maiorano L, Falcucci A, Garton EO, Boitani L (2007) Contribu-
tion of the Natura 2000 Network to Biodiversity Conservation
in Italy. Conservation Biology,21, 1433–1444.
Maret TJ, Snyder JD, Collins JP (2006) Altered drying regime
controls distribution of endangered salamanders and intro-
duced predators. Biological Conservation,127, 129–138.
McKinney ML (2006) Urbanization as a major cause of biotic
homogenization. Biological Conservation,127, 247–260.
McKinney ML, Lockwood JL (1999) Biotic homogenization: a
few winners replacing many losers in the next mass extinction.
Trends in Ecology and Evolution,14, 450–453.
Nakicenovic N, Swart R (2000) Emissions Scenarios: A Special
Report of Working Group III of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cambridge.
Nielsen C, Hartvig P, Kollmann J (2008) Predicting the distribu-
tion of the invasive alien Heracleum mantegazzianum at two
different spatial scales. Diversity and Distributions,14, 307–317.
Olden JD, Kennard MJ, Pusey BJ (2008) Species invasions and the
changing biogeography of Australian freshwater fishes. Global
Ecology and Biogeography,17, 25–37.
Olden JD, Poff NL, Douglas MR, Douglas ME, Fausch KD (2004)
Ecological and evolutionary consequences of biotic homoge-
nization. Trends in Ecology and Evolution,19, 18–24.
Pearson RG, Raxworthy CJ, Nakamura M, Peterson AT (2007)
Predicting species distributions from small numbers of occur-
rence records: a test case using cryptic geckos in Madagascar.
Journal of Biogeography,34, 102–117.
Petit CC, Lambin EF (2002) Long-term land-cover changes in the
Belgian Ardennes (1775–1929): model-based reconstruction vs.
historical maps. Global Change Biology,8, 616–630.
Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy
modeling of species geographic distributions. Ecological Mod-
elling,190, 231–259.
Phillips SJ, Dudı
´k M (2008) Modeling of species distributions
with Maxent: new extensions and comprehensive evaluation.
Ecography,31, 161–175.
Quian H, Ricklefs RE (2006) The role of exotic species in
homogenizing the North American flora. Ecology Letters,9,
Ricciardi A (2007) Are modern biological invasions an unprece-
dented form of global change? Conservation Biology,21, 329–336.
Rounsevell MDA, Reginster I, Araujo MB et al. (2006) A coherent
set of future land use change scenarios for Europe. Agriculture
Ecosystems and Environment,114, 57–68.
Roura-Pascual N, Suarez AV, Goomez C, Pons P, Touyama Y,
Wild AL, Peterson AT (2004) Geographical potential of Argen-
tine ants (Linepithema humile Mayr) in the face of global climate
change. Proceedings of the Royal Society of London B: Biological
Sciences,271, 2527–2534.
Sala OE, Chapin FS, Armesto JJ et al. (2000) Biodiversity – global
biodiversity scenarios for the year 2100. Science,287,
Societas Herpetologica Italica. (2008) Osservatorio Erpetologico
Italiano. Acta Herpetologica,3, 83–84.
Spangenberg JH (2007) Integrated scenarios for assessing biodi-
versity risks. Sustainable Development,15, 343–356.
Strayer DL, Evinver VT, Jeschke JM, Pace ML (2006) Under-
standing the long-term effects of species invasions. Trends in
Ecology and Evolution,21, 645–651.
Thuiller W (2007) Climate change and the ecologist. Nature,448,
Thuiller W, Richardson DM, Pysek P, Midgley GF, Hughes GO,
Rouget M (2005) Niche-based modelling as a tool for predict-
ing the risk of alien plant invasions at a global scale. Global
Change Biology,11, 2234–2250.
Urban MC, Phillips BL, Skelly DK, Shine R (2008) A toad more
traveled: the heterogeneous invasion dynamics of cane toads
in Australia. The American Naturalist,171, E134–E148.
Vallecio S, Brotons L, Thuiller W (2009) Dangers of predicting
bird species distributions in response to land-cover changes:
the role of dynamic processes. Ecological Applications,19,
Ward DF (2007) Modelling the potential geographic distribution
of invasive ant species in New Zealand. Biological Invasions,9,
Wisz MS, Hijmans RJ, Li J, Peterson AT, Graham CH, Guisan A
(2008) Effects of sample size on the performance of
species distribution models. Diversity and Distributions,14,
Yiming L, Zhengjun W, Duncan RP (2006) Why islands are easier
to invade: human influences on bullfrog invasion in the
Zhoushan archipelago and neighboring mainland China.
Oecologia,148, 129–136.
r2009 Blackwell Publishing Ltd, Global Change Biology,16, 528–537
... In Arizona, K. s. sonoriense (hereafter, desert mud turtle) is adversely affected by invasive species, particularly by two species of crayfish (Faxonius virilis and Procambarus clarkii) and the American bullfrog (Lithobates catesbeianus), which originated from eastern North America and are widely distributed in freshwater ecosystems across the south-western USA (Rosen & Schwalbe, 1995;Carpenter, 2005;Hensley et al., 2010). These three species are highly successful generalists that have invaded freshwater systems globally and exert adverse impacts on native biota by altering ecosystem dynamics and trophic interactions (Adams & Pearl, 2007;Ficetola, Thuiller & Miaud, 2007;Gherardi, 2007;Gherardi et al., 2011). In addition to indirect bottom-up effects, all three of these invasive species prey upon hatchling mud turtles and decrease population recruitment (Akins & Jones, 2010;Hensley et al., 2010). ...
... In this study, higher abundance of bullfrogs and crayfish were detected at higher elevations (even above 1,000 masl), which might be explained by an indirect association of higher elevations with suitable conditions for these species, such as other climatic, physicochemical or biological parameters (e.g. temperature; Seiler & Turner, 2004;Ficetola, Thuiller & Miaud, 2007;Adams & Marks, 2016). ...
... Other studies have also reported higher invasive abundance in more disturbed environments Ficetola, Thuiller & Miaud, 2007;Sepúlveda et al., 2015;Sepulveda, 2018). In contrast, crayfish abundance was significantly positively correlated with distance from cities, where lower abundances were recorded closer to cities. ...
Identifying the ecological factors that determine the spread of invasive species is key to adequately managing endangered species in freshwater ecosystems. Invasive species are a main threat to turtles, which are targets of major conservation efforts worldwide. In freshwater ecosystems of the south‐western USA, invasive bullfrog (Lithobates catesbeianus) and crayfish species (Faxonius virilis and Procambarus clarkii) represent a major risk to the desert mud turtle (Kinosternon sonoriense sonoriense), state‐listed as a Species of Greatest Conservation Need in Arizona. As a species in the early stages of population decline, the desert mud turtle is a top candidate for the development of management plans to decrease extinction risk. An invasion risk assessment tool was built from available occurrence data for K. s. sonoriense and the invasive bullfrog and crayfish species in Arizona using 5,886 de‐duplicated records from public databases and reports from the Arizona Game and Fish Department. The occurrence density of K. s. sonoriense was calculated state‐wide to define populations in which the level of invasion by bullfrog and crayfish was assessed. The environmental factors associated with the abundance of invasive species in populations of K. s. sonoriense were then analysed. A higher prevalence of crayfish and bullfrog was detected in turtle populations located in perennial streams. Invasive abundance was significantly higher in turtle populations at higher elevation and closer to the main river trunk for both invasive taxa. Higher bullfrog abundance was detected near human settlements, whereas crayfish were more abundant further from human settlements. These results will inform which populations of K. s. sonoriense require intensive surveying and control of invasive species to maintain the health of native desert mud turtle populations. This study provides valuable information regarding the environmental conditions associated with the abundance of invasive species threatening turtle populations, helping to develop science‐based management of freshwater ecosystems.
... Information on introduced bullfrog population genetics is crucial to understand invasion history, structure and dynamics of gene pool exchange between populations 47,48 , and can be helpful to develop management and control programs 49,50 . Efforts to understand the genetic structure of introduced bullfrog populations have been undertaken in several regions, such as Europe 51 , China 43 and western USA 48,52 . ...
... The same process seems to have occurred in other countries where populations were analyzed with the same genetic markers used in this study. Similar patterns were observed in China 43 -where also only two haplotypes were found among over 500 samples-and in some degree in Europe (5 haplotypes among nearly 400 samples from 8 countries) 47 , although European countries have more populations and more genetic diversity, due to the relatively high number of introduction events that happened in the continent (at least 25) 51 . The process of transforming introduced populations into a single population due to exchange of migrants may obscure population and invasion genetics inference, as the genetic signal is lost making it nearly impossible to clarify the history of events that followed the introductions. ...
... The number of introduction events seems to be directly related to the degree of genetic diversity in the region, and the bullfrog populations in Brazil exhibit the lowest number of mtDNA haplotypes of all studied non-native populations of this species examined so far 43,47,52 . Although the genetic diversity of bullfrogs in Brazil seems to be low, especially when compared to populations in its native range (42 haplotypes were found in the United States 79 ), farming does not seem to suffer any impacts. ...
Full-text available
Non-native species are a major problem affecting numerous biomes around the globe. Information on their population genetics is crucial for understanding their invasion history and dynamics. We evaluated the population structure of the non-native American bullfrog, Aquarana catesbeiana, in Brazil on the basis of 324 samples collected from feral and captive groups at 38 sites in seven of the nine states where feral populations occur. We genotyped all samples using previously developed, highly polymorphic microsatellite loci and performed a discriminant analysis of principal components together with Jost’s D index to quantify pairwise differentiation between populations. We then amplified 1,047 base pairs of the mitochondrial cytochrome b (cytb) gene from the most divergent samples from each genetic population and calculated their pairwise differences. Both the microsatellite and cytb data indicated that bullfrogs comprise two populations. Population grouping 1 is widespread and possesses two cytb haplotypes. Population grouping 2 is restricted to only one state and possesses only one of the haplotypes from Population grouping 1. We show that there were two imports of bullfrogs to Brazil and that there is low genetic exchange between population groupings. Also, we find that there is no genetic divergence among feral and captive populations suggesting continuous releases. The limited genetic variability present in the country is associated to the small number of introductions and founders. Feral bullfrogs are highly associated to leaks from farms, and control measures should focus on preventing escapes using other resources than genetics, as feral and captive populations do not differ.
... A causa de las repetidas introducciones para acuicultura, han sido reportadas poblaciones asilvestradas de rana toro en diversos sitios de Asia, Europa y América (Ficetola et al. 2007, Cunha & Delariva 2009, Kraus 2009). En Sudamérica, Brasil es el país que exhibe el mayor número de focos (Both et al. 2011), pero también en Argentina, Uruguay, Ecuador, Venezuela y Colombia existen diversas poblaciones de esta especie exótica invasora. ...
... Diferentes estudios de modelos de nicho predicen condiciones óptimas para la expansión de L. catesbeianus en prácticamente todo el territorio uruguayo (e.g., Ficetola et al. 2007, Nori et al. 2011. A escala de paisaje, las características del terreno y la disponibilidad de cuerpos de agua lénticos condicionan el avance de esta especie. ...
Full-text available
La rana toro Lithobates catesbeianus, originaria del este de Norteamérica, fue introducida en Uruguay en la década de 1980, para la cría en granjas por su uso comestible. Así llegaron a funcionar 19 granjas, que en dos décadas terminaron cerrando sin un control del destino de este anuro acuático de gran tamaño corporal, que actúa como un depredador tope invasor y vector de enfermedades. El Proyecto Rana Toro en Uruguay, monitorea desde 2005 (primer registro) la evolución de esta invasión y sus efectos. Hasta la fecha en Uruguay se detectaron cinco poblaciones, de las cuales persisten las de Aceguá (Cerro Largo), San Carlos (Maldonado) y Los Cerrillos (Canelones). En Aceguá la población comenzó su expansión en 2012, ocupando actualmente un área de 4.4 km2. Si bien el conocimiento del proceso resulta menor, el área invadida en San Carlos es mayor (6.4 km2). Finalmente, la situación de la población recientemente detectada en Los Cerrillos es la menos conocida. Los estudios de campo en dichas localidades muestran que las comunidades invadidas se caracterizan por un decrecimiento en la riqueza y abundancia de anuros nativos y por un aumento en las densidades y tamaño corporal de las mojarras. La rana toro es un fuerte estructurador de las comunidades invadidas. Por esto, se ha trabajado el componente social, promoviendo la comprensión y participación ciudadana en los planes de manejo. Simultáneamente se trabajó con las autoridades ambientales, quienes en 2012 declaran prioritario el control de esta invasora. Actualmente existe una intención de aplicar medidas de control por parte de las autoridades, cuyo éxito dependerá de la capacitación de sus técnicos, de su capacidad de involucrar a los actores locales y considerar el conocimiento científico. El control y la erradicación son aún posibilidades reales si se consigue una pronta, fuerte y eficaz respuesta.
... A causa de las repetidas introducciones para acuicultura, han sido reportadas poblaciones asilvestradas de rana toro en diversos sitios de Asia, Europa y América (Ficetola et al. 2007, Cunha & Delariva 2009, Kraus 2009). En Sudamérica, Brasil es el país que exhibe el mayor número de focos (Both et al. 2011), pero también en Argentina, Uruguay, Ecuador, Venezuela y Colombia existen diversas poblaciones de esta especie exótica invasora. ...
... Diferentes estudios de modelos de nicho predicen condiciones óptimas para la expansión de L. catesbeianus en prácticamente todo el territorio uruguayo (e.g., Ficetola et al. 2007, Nori et al. 2011. A escala de paisaje, las características del terreno y la disponibilidad de cuerpos de agua lénticos condicionan el avance de esta especie. ...
Full-text available
El ligustro, Ligustrum lucidum W.T. Aiton (Oleaceae), es una de las principales especies exóticas invasoras de los bosques de Uruguay. Su alta producción de semillas y banco de plántulas, dispersión por aves frugívoras, rápido crecimiento y amplia tolerancia ambiental, le confieren un gran potencial invasor. Este árbol originario de Asia templada (principalmente China) ha invadido actualmente todos los continentes, excepto la Antártida. En Uruguay, fue introducido a mediados del siglo XIX como especie ornamental, para ser usado en parques, plazas y cercos vivos. En la actualidad, se ha registrado su presencia en el 4.3 % de las 1467 parcelas relevadas por el Inventario Forestal Nacional, principalmente en bosques ribereños y parques. Se distribuye principalmente en el sur, litoral oeste y centro del país. Por el momento, los registros en el norte y este son escasos. La invasión del ligustro genera impactos sobre la diversidad de los bosques, alterando su estructura y funcionamiento, pudiendo llegar a extinguir localmente a algunas especies leñosas. Si bien la superficie de bosques invadidos no llegaría al 5% en la actualidad, debido a la gran extensión geográfica (14-15 departamentos) y alto potencial invasor, la erradicación del ligustro es prácticamente imposible a nivel nacional. Una estrategia razonable sería enfocarse en cuatro tipos de acciones: (1) Reducción de fuentes de propágulos, mediante la prohibición de venta de ligustros en viveros y el control en centros poblados (cercos vivos, plazas). (2) Monitoreo y prevención, para erradicar invasiones recientes, principalmente en las regiones norte y este del país. (3) Control para reducir abundancia en áreas de relevancia ecológica que cuenten con recursos para su gestión a largo plazo, como es el caso de áreas protegidas públicas o privadas. (4) Investigación sobre ecología de la invasión y métodos de restauración de bosques invadidos.
... A causa de las repetidas introducciones para acuicultura, han sido reportadas poblaciones asilvestradas de rana toro en diversos sitios de Asia, Europa y América (Ficetola et al. 2007, Cunha & Delariva 2009, Kraus 2009). En Sudamérica, Brasil es el país que exhibe el mayor número de focos (Both et al. 2011), pero también en Argentina, Uruguay, Ecuador, Venezuela y Colombia existen diversas poblaciones de esta especie exótica invasora. ...
... Diferentes estudios de modelos de nicho predicen condiciones óptimas para la expansión de L. catesbeianus en prácticamente todo el territorio uruguayo (e.g., Ficetola et al. 2007, Nori et al. 2011. A escala de paisaje, las características del terreno y la disponibilidad de cuerpos de agua lénticos condicionan el avance de esta especie. ...
Full-text available
El caracol rapana, Rapana venosa (Valenciennes, 1846) es un predador activo de más de nueve especies nativas de moluscos que ha llegado a aguas del Río de la Plata en 1998 desde el sudeste asiático, posiblemente en aguas de lastre. Desde su llegada su distribución se ha ido ampliando hacia el sur por la costa de la provincia de Buenos Aires y hacia el norte por la costa uruguaya. En la actualidad parece estar desplazándose, ampliando su distribución, hacia la zona Sur de la costa argentina. El presente capítulo describe resultados de los estudios desarrollados para el estuario del Río de la Plata con énfasis en aspectos de distribución, y en los efectos ecosistémicos que puede ocasionar. Se presentan las acciones de control desarrolladas en Uruguay y Argentina hasta el momento y se sugiere que el uso como recurso alimenticio explotable podría tener viabilidad. Se destaca el elevado valor comercial como ítem exportable y el desarrollo de importantes pesquerías en aquellos lugares donde es consumido. De acuerdo a los resultados presentados, claramente rapana representa un riesgo para la malacofauna en general y para la de importancia económica de la zona, en especial para el mayor recurso malacológico explotado en Uruguay, el mejillón azul. También ocurre competencia, por el espacio, con especies nativas y causa impacto directo sobre la tortuga verde a través del biofouling sobre el carapacho, generando problemas de flotabilidad y nado. Se recomienda, entre otras acciones, la promoción de su captura como recurso pesquero sobre-explotable y su potencial uso como indicador de calidad ambiental en el Río de la Plata y zona oceánica adyacente. Palabras clave: caracol rapana, especie invasora, estuario, mitigación y control, predación
... A causa de las repetidas introducciones para acuicultura, han sido reportadas poblaciones asilvestradas de rana toro en diversos sitios de Asia, Europa y América (Ficetola et al. 2007, Cunha & Delariva 2009, Kraus 2009). En Sudamérica, Brasil es el país que exhibe el mayor número de focos (Both et al. 2011), pero también en Argentina, Uruguay, Ecuador, Venezuela y Colombia existen diversas poblaciones de esta especie exótica invasora. ...
... Diferentes estudios de modelos de nicho predicen condiciones óptimas para la expansión de L. catesbeianus en prácticamente todo el territorio uruguayo (e.g., Ficetola et al. 2007, Nori et al. 2011. A escala de paisaje, las características del terreno y la disponibilidad de cuerpos de agua lénticos condicionan el avance de esta especie. ...
Full-text available
CAPÍTULO VIII. Limnoperna fortunei (Mejillón dorado): características bióticas, distribución, impactos y manejo poblacional en Uruguay Ernesto Brugnoli, Jennifer Pereira, Carolina Ferrer, Ivana Silva, Leandro Capurro, Ana Laura Machado, Juan María Clemente (†), Lucía Boccardi, Soledad Marroni, Daniel Fabián, Fabiana Rey, María Jesús Dabezies, Iván González-Bergonzoni, Daniel Naya, Alejandro D’Anatro, Franco Teixeira de Mello, Claudio Martínez, Guillermo Goyenola, Carlos Iglesias & Pablo Muniz. Resumen El mejillón dorado (Limnoperna fortunei) es una especie de molusco originario de Asia que ingresó a inicios de 1990 al Río de la Plata por medio de aguas de lastre; actualmente es considerada como especie invasora en la cuenca del Plata. Presenta hábitos bentónicos, epifaunal bisado, comportamiento gregario, desarrollo indirecto, presencia de estadios larvales y ciclo reproductivo asociado a la variación de la temperatura del agua. Invade los principales cuerpos hídricos de la región (Argentina, Brasil, Paraguay, Bolivia y Uruguay). En nuestro país se reporta en sistemas hídricos de las cuencas de los ríos Uruguay, Negro, Santa Lucía, zonas interna y media del Río de la Plata, Laguna Merín y Laguna del Sauce, presentando diferencias en las tasas de invasión a nivel regional y local. En la región se reporta asociado a sustratos consolidados naturales y artificiales, incrementado sus abundancias poblacionales, ocasionando modificaciones en las comunidades bentónicas y planctónicas, en hábitos alimenticios de peces autóctonos y modificaciones en los parámetros de calidad de agua. Se presentan estudios desarrollados en Uruguay sobre efectos ecosistémicos y usos de la especie como bio-monitor de calidad de agua. El mejillón dorado ocasiona efectos de macrofouling (incrustaciones) adhiriéndose a infraestructuras hidráulicas, generando problemas en servicios ecosistémicos en diferentes usuarios de los recursos hídricos de la cuenca del Plata. Se describen generalidades sobre estrategias de manejo poblacional desarrolladas para la prevención, control/mitigación y erradicación de especies invasoras que ocasionan macrofouling mostrando ejemplos nacionales sobre prevención y control (mecánico, pinturas antiincrustantes, biológico) realizados en estudios de grado, posgrado y proyectos de investigación para el manejo poblacional de la especie. Palabras claves: efectos ecosistémicos, especie acuática invasora, macrofouling, manejo poblacional, servicios ecosistémicos.
... In addition, it has been often used to describe potential areas of distribution where rare species could occupy (e.g., Raxworthy 2003;Giovanelli et al. 2008a; Barbosa et al. 2015). Furthermore, this approach has also been used to predict the potential entry areas/invasion routes of non-indigenous species (e.g., Vieglais and Peterson 2001;Papes and Peterson 2003;Peterson et al. 2003;Ficetola et al. 2007;Thapa et al. 2018). ...
The genus Baccharis (Asteraceae) comprises over 440 species distributed along North and South America. Some species of this genus have remarkable invasiveness, and one of these species is the South American shrub Baccharis dracunculifolia DC. Most of the introductions of non-indigenous species are held indirectly through trade, so it is believed that this species could become invasive worldwide with a particular interest in the North American continent because of the increasing sale of products derived from honey to this continent. The resin extracted from B. dracunculifolia is the leading source for preparing the green propolis produced in Southeastern Brazil. Thus, the main objective of this work is to apply an approach based on distribution modeling to investigate possible areas of high environmental suitability for B. dracunculifolia in the North American continent and the potential to the entire globe using current and two future scenaries. Our results show many areas of environmental suitability for B. dracunculifolia. This species can invade over 33 countries distributed into five continents, including North America, some of the most critical parts of the southern USA, and large areas in Mexico. Since the best way of countering biological invasions is prevention, we propose that the introduction of this species should be monitored.
... Overall, our research adds to growing evidence that nonnative American Bullfrogs have limited effects on amphibians in lowland systems in the Pacific Northwest, particularly in comparison to the effect of nonnative fish or declining habitat quality Jennings 1986, Adams et al. 1998). However, it should be noted that adult and larval American Bullfrogs are considered harmful nuisance species in other areas of the country (e.g., southwestern USA; Mims et al. 2020) and the world (Ficetola et al. 2007), and stable isotopes may be useful in examining trophic shifts in those systems. ...
Full-text available
The invasive species are of global concern, and the Invasive American Bullfrog (IAB; Lithobates catesbeianus) is one of the worst invasive amphibian species worldwide. Like other countries, South Korea is also facing challenges from IAB. Although many studies indicated impacts of IAB on native anurans in Korea, the actual risk at the specific level is yet to evaluate. Considering the putative invasiveness of IAB, it is hypothesized that any species with the possibility of physical contact or habitat sharing with them, will have a potential risk. Thus, we estimated and observed their home range, preferred habitats, morphology, behavior, and ecology. Then, comparing with existing knowledge, we assessed risks to the native anurans. We found a home range of 3474.2 ± 5872.5 m² and identified three types of habitats for IAB. The analyses showed at least 84% of native anurans (frogs and toads) were at moderate to extreme risks, which included all frogs but only 33% of toads. Finally, we recommended immediate actions to conserve the native anurans based on our results. As this study is the first initiative to assess the specific risk level from the invasiveness of L. catesbeianus, it will help the managers to set conservation priorities and strategies.
Full-text available
Determining the invasibility of habitats by alien species is crucial for understanding their spread potential, the habitats the most at risk and to implement adequate management actions. This is urgent for introduced taxa that show high invasion potential across broad geographical scales. We here assess these processes in invasive Pelophylax water frogs which are widespread colonizers across Western Europe and for which the invasibility of habitats remains to be quantified. Specifically, we used hierarchical occupancy models in a Bayesian framework to identify local- and landscape-scale features that can enhance occupancy of the most common invasive water frog, the marsh frog (P.ridibundus), in southern France. Water frogs were highly detectable and showed high occupancy across the invaded landscape. The invaders expressed a very broad habitat tolerance for both local- and landscape-scale variables while their invasion was facilitated by the occurrence of deep, permanent ponds with abundant aquatic vegetation and high sun exposure. Cross-validation showed a good transferability of models across space. The high invasibility of a wide range of habitats by Pelophylax water frogs is alarming and unveils their invasiveness, contributing therefore to explain their success of invasion over broad geographic scales.
Full-text available
Full-text available
The evidence for rapid climate change now seems overwhelming. Global temperatures are predicted to rise by up to 4 °C by 2100, with associated alterations in precipitation patterns. Assessing the consequences for biodiversity, and how they might be mitigated, is a Grand Challenge in ecology.
Predicting the probability of successful establishment of plant species by matching climatic variables has considerable potential for incorporation in early warning systems for the management of biological invasions. We select South Africa as a model source area of invasions worldwide because it is an important exporter of plant species to other parts of the world because of the huge international demand for indigenous flora from this biodiversity hotspot. We first mapped the five ecoregions that occur both in South Africa and other parts of the world, but the very coarse definition of the ecoregions led to unreliable results in terms of predicting invasible areas. We then determined the bioclimatic features of South Africa's major terrestrial biomes and projected the potential distribution of analogous areas throughout the world. This approach is much more powerful, but depends strongly on how particular biomes are defined in donor countries. Finally, we developed bioclimatic niche models for 96 plant taxa (species and subspecies) endemic to South Africa and invasive elsewhere, and projected these globally after successfully evaluating model projections specifically for three well-known invasive species (Carpobrotus edulis, Senecio glastifolius, Vellereophyton dealbatum) in different target areas. Cumulative probabilities of climatic suitability show that high-risk regions are spatially limited globally but that these closely match hotspots of plant biodiversity. These probabilities are significantly correlated with the number of recorded invasive species from South Africa in natural areas, emphasizing the pivotal role of climate in defining invasion potential. Accounting for potential transfer vectors (trade and tourism) significantly adds to the explanatory power of climate suitability as an index of invasibility. The close match that we found between the climatic component of the ecological habitat suitability and the current pattern of occurrence of South Africa alien species in other parts of the world is encouraging. If species' distribution data in the donor country are available, climatic niche modelling offers a powerful tool for efficient and unbiased first-step screening. Given that eradication of an established invasive species is extremely difficult and expensive, areas identified as potential new sites should be monitored and quarantine measures should be adopted.
Bullfrogs maintain a homeostasis of body temperature comparable to that of many reptiles. Body temperatures were measured immediately after capturing animals in the field and by continuous monitoring of free-ranging adults containing miniature radio transmitters. The range of body temperature during normal diurnal activity was 26-33° C, and the mean was approximately 30° C. The critical thermal maximum was 38.2° C. Body temperatures corresponded more closely to dry bulb than to "wet bulb" air temperatures. Behavioral thermoregulation is achieved by changes in location and alterations in posture. Radiant energy is used as a heat source, and pond water is used either as a heat source or as a heat sink. Cooling is augmented by postural adjustments which increase the amount of surface area exposed to evaporation. Postural thermoregulation was observed more frequently in young bullfrogs (which often occupy the warmest microenvironments available) than in adults. Temperatures of the surface and shallow waters of small ponds may influence frogs to emerge onto the bank where they are exposed to insolation. While out of water bullfrogs may heat, cool, or maintain a relatively constant body temperature depending on prevailing environmental conditions and related postural adjustments.