Management of Biological Invasions (2019) Volume 10, Issue 3: 428–448

CORRECTED PROOF

Research Article Between the hammer and the anvil: how the combined effect of global warming and the non-native common slider could threaten the European pond turtle

Francesco Cerasoli, Mattia Iannella* and Maurizio Biondi University of L’Aquila, Department of Life, Health & Environmental Sciences, Via Vetoio snc, 67100 L'Aquila – Coppito, Italy Author e-mails: [email protected] (FC), [email protected] (MI), [email protected] (MB) *Corresponding author

Citation: Cerasoli F, Iannella M, Biondi M (2019) Between the hammer and the anvil: Abstract how the combined effect of global warming and the non-native common The European pond turtle Emys orbicularis is seriously threatened across Europe slider could threaten the European pond by the non-native common slider, Trachemys scripta, which is included among the turtle. Management of Biological Invasions 100 most invasive species worldwide. Using ensemble forecast techniques, we 10(3): 428–448, https://doi.org/10.3391/mbi. analysed the influence of climatic factors on the current distribution of these two 2019.10.3.02 species, subsequently projecting the obtained models under two Representative Received: 7 May 2018 Concentration Pathways (RCPs) to 2050 and 2070, to investigate if global warming Accepted: 2 April 2019 would boost direct competition within our study region. We implemented a gap Published: 27 May 2019 analysis in GIS environment to assess how protected areas (PAs) may be affected by Handling editor: Amy Yackel a loss of suitable area for E. orbicularis and by an expansion of T. scripta. An analysis Thematic editor: Catherine Jarnevich of altitudinal range shift was evaluated, based on obtained projections. We found that both species may gain suitable area in the future, possibly because of the positive Copyright: © Cerasoli et al. effect of increasing temperatures, and are predicted to shift from plain-to-hilly areas This is an open access article distributed under terms of the Creative Commons Attribution License towards higher altitudes. These trends result in an increasing overlap of potentially (Attribution 4.0 International - CC BY 4.0). suitable areas for both species, particularly within PAs; moreover, a niche analysis OPEN ACCESS. highlights that the results obtained are linked to the environmental niches of the two species. Our findings suggest the necessity, particularly for PAs’ authorities, of field monitoring of T. scripta and further research to more comprehensively assess biotic and abiotic factors influencing the invasiveness of this species.

Key words: ensemble forecast, climate change, Trachemys scripta, gap analysis, altitudinal shift, GIS analysis, Emys orbicularis

Introduction The implementation of Species Distribution Models (SDMs) in the analytical framework aiming at establishing optimal practices for conservation purposes has increased in recent years (Loiselle et al. 2003; Urbina-Cardona and Flores-Villela 2010; Ferreira et al. 2013; Guisan et al. 2013; Treglia et al. 2015; Urbani et al. 2015; Cerasoli et al. 2017; Iannella et al. 2018a, b). A particularly interesting branch of this research field focuses on the application of SDMs to investigate the expansion dynamics of invasive alien species (IAS) to better inform mitigation measures (Peterson 2003; Ficetola et al. 2007, 2009; Rodder et al. 2009; van Wilgen and Richardson 2012). Indeed, the successful establishment of IAS is one of the most relevant issues in conservation biology because of the consequent significant loss of

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 428 Invasive common slider and global warming threaten European pond turtle

native species and ecosystem services, with the associated management problems and economic costs (e.g. Oreska and Aldridge 2011; Caffrey et al. 2014; Gilioli et al. 2017) observed during recent decades. The diffusion of invasive species worldwide has not slowed during recent years (Seebens et al. 2017), despite noticeable efforts in studying, monitoring, controlling and eradicating several taxa by international programs (e.g. DAISIE project for the EU, the NISC for the U.S.A. and the CABI, ISSG or the Convention on Biological Diversity). The common slider, Trachemys scripta (Thunberg in Schoepff, 1792), is considered one of the 100 worst IAS in Europe (Scalera 2009) and the most invasive reptile in the world (Kraus 2009); imported for pet-trade from North America (Telecky 2001), this species has been repeatedly released in nature, until the import was stopped by the EU Wildlife Trade Regulation (European Union 1997). However, this act only involved T. scripta elegans (Wied, 1839), but in many countries it was not extended to the other two subspecies, T. scripta scripta and T. scripta troosti (Holbrook, 1836), which have flooded the reptile pet market, with the subsequent release of specimens in the wild (EEA 2012). The common slider is proven to cause several problems to the European pond turtle, Emys orbicularis (Linnaeus, 1758), a species of conservation concern in Italy. Land use change and habitat loss are certain causes of the decline of E. orbicularis (TFTS Group 1996), but the presence of T. scripta, which usually represents a tough competitor of the former species, exacerbate the problems for its protection (Polo-Cavia et al. 2014). In fact, T. scripta can successfully compete with E. orbicularis for basking places (Cadi and Joly 2003), causing weight loss and high mortality to this latter species (Cadi and Joly 2004). Furthermore, T. scripta is a vector of both non-native (Oi et al. 2012; Meyer et al. 2015) and native parasites, cross transmitting them among different native turtle populations (Hidalgo-Vila et al. 2009). Propagule pressure, defined as the disturbance produced by non-natives released in a non-native area, together with resource availability in terms of food, living space and habitat, represent important variables in determining a possible IAS establishment (Kraus 2009; Berg et al. 2016). Along with these factors primarily acting at local scales, IAS establishment can be prompted also by climate change (Occhipinti-Ambrogi 2007; Walther et al. 2009; Bomford et al. 2010; Bellard et al. 2013; Sorte et al. 2013). Considering future climate-change scenarios, the invasion of T. scripta and the possible shift of suitable areas for E. orbicularis could follow

uncertain trajectories linked to CO2 increase. Indeed, different trajectories

of atmospheric CO2 concentration are predicted for the next decades by several Global Climate Models (GCMs), resulting from the Coupled Model Intercomparison Project Phase 5 (CMIP5) (Meehl et al. 2009), each of which depicts a different future climate scenario. Depending on the climate- change scenario considered, various dynamics in ecosystem invasions by

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 429 Invasive common slider and global warming threaten European pond turtle

non-native species can be inferred; in fact, this variation among the predictions of different Global Climate Models (GCM) has been recognised as one of the main sources of uncertainty for SDMs projection to future scenarios (Thuiller et al. 2004; Garcia et al. 2012; Stralberg et al. 2015). Our research evaluates the current and future potential distribution of T. scripta in the Italian peninsula and Sardinia with respect to E. orbicularis, based on Ficetola et al. (2009). We add new data and implement a modelling framework taking advantage of ensemble techniques; indeed, recent years have seen a drastic increase both in the development of platforms permitting species distribution modelling (e.g. Naimi and Araújo 2016; Thuiller et al. 2016) and in papers debating and/or applying it (Buisson et al. 2010; Stohlgren et al. 2010; Grenouillet et al. 2011; Poulos et al. 2012; Zhang et al. 2015).

Materials and methods Species and study region Emys orbicularis, classified as “Near threatened” by the IUCN (TFTS Group 1996) and as “Endangered – A2c” in Italy (Andreone et al. 2013), lives in different types of aquatic habitats, ranging from lakes and ponds to riversides and ditches (Zuffi et al. 2011). Trachemys scripta has rapidly colonized habitats previously occupied by E. orbicularis (Di Tizio and Di Cerbo 2011). A GPS-resolution database of presence points comprising 314 records for E. orbicularis and 4008 (3757 for native and 251 for invaded range) records for T. scripta (see Supplementary material Table S1) was generated through the integration of literature data, online search and unpublished data (see Figure 1 for presence points falling within the study region and Figure S1 for native presence localities). Some of the occurrence data for T. scripta falling within the invaded range are also reproductive sites, which could be informative with respect to the fitness of an IAS (Ficetola et al. 2009). Nonetheless, we did not consider these sites as a separate data set for modelling for two reasons: the records mostly corresponded to sites already known and considered in Ficetola et al. (2009) and the scarce number of records (< 20) would not properly calibrate the SDMs. Our study region includes the Italian peninsula and Sardinia (Figure 1), where both E. orbicularis and the invasive T. scripta occur. In order to highlight areas of possible current co-occurrence of the two species, we calculated in ArcMap 10.0 (ESRI 2010) the number of occurrence localities for E. orbicularis with at least one T. scripta record falling within 10 km and 1 km radius (based on T. scripta range and mean movements reported in Burke et al. (1995) and in Ryan et al. (2008)).

Model building Worldclim (ver. 1.4) bioclimatic variables at 30” resolution were chosen as candidate predictors (Hijmans et al. 2005) for both the current and the future

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 430 Invasive common slider and global warming threaten European pond turtle

Figure 1. Physical map of the Italian peninsula and Sardinia region depicting the distribution of the two turtle species. Occurrences of Trachemys scripta are shown as black triangles, while Emys orbicularis is represented by pink circles with black dots.

scenarios (Appendix 1). In order to investigate if and how suitability for E. orbicularis and T. scripta within the study region would change in the future, two “Representative Concentration Pathways” (RCPs) (Meinshausen et al. 2011; Stocker et al. 2014) were chosen to project the SDMs to the 2050 and 2070 scenarios: RCP4.5, as application of the COP21 commitments, in which a slight and controlled increase in greenhouse gases with a peak in 2040 is foreseen, and RCP8.5, also known as the “business as usual” scenario, with the emissions continuing to rise throughout the 21st century (Riahi et al. 2011). Following a frequently used approach for dealing with GCM differences (Khanum et al. 2013; Ashraf et al. 2017; Zhang et al. 2017), we chose to project the obtained SDMs to the conditions predicted by four GCMs, namely BCC-CSM-1 (Wu et al. 2014), CCSM4 (Gent et al. 2011), CNRM-CM5 (Voldoire et al. 2011) and MIROC-ESM (Watanabe et al. 2011), available on the Worldclim repository. Trachemys scripta models were calibrated over both the native and the invaded range, to improve prediction performance and reduce model uncertainty (Broennimann and Guisan 2008).

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 431 Invasive common slider and global warming threaten European pond turtle

In order to avoid spatial autocorrelation among presence records, which could result in misleading predictions (Fourcade et al. 2014) and inflation of model discrimination metrics (Veloz 2009), the database for each species was spatially thinned through the spThin R package (thin.par = 10 km, reps = 10) (Aiello‐Lammens et al. 2015). The thinned records were further tested for spatial autocorrelation through the Moran’s I test in ArcMap 10.0 (ESRI 2010). Multicollinearity within our set of 19 bioclimatic variables can cause distortions in the estimation of model parameters and variable importance (Dormann 2007; Crase et al. 2012) and was addressed by building a correlation matrix (Table S2). For the pairs of variables with Pearson | r | > 0.85 (Elith et al. 2006), we chose to keep the variable known to be more relevant in terms of the species’ life-history traits (Ficetola et al. 2009; Di Tizio and Di Cerbo 2011; Zuffi et al. 2011; Dormann et al. 2013; Brandt et al. 2017). When a pair of variables, both of which are considered biologically relevant, slightly exceeded the selected threshold of r, we kept both as predictors for model building (Rissler and Apodaca 2007; Werneck et al. 2011). The presence-background SDMs (Guillera-Arroita et al. 2015) for the two species were built in R (R Core Team 2016) through the ensemble modelling package biomod2 (Thuiller et al. 2016). For the current conditions, the “BIOMOD_FormatingData” function was used to generate 10 sets of 1000 pseudo-absences each. Disk strategy (Thuiller et al. 2016) was chosen to randomly select the pseudo-absences within a buffer of 2 ÷ 150 km for E. orbicularis and 2 ÷ 600 km for T. scripta; this last buffer is wider because of the greater extent of T. scripta native range. The algorithms used for model building were: a) parametric Generalized Linear Models (GLM) (McCullagh 1984); b) spline-based regression Multivariate Adaptive Regression Splines (MARS) (Leathwick et al. 2005); the two machine learning techniques c) Gradient Boosting Models (GBM) (Elith et al. 2008) and d) Maxent (Phillips et al. 2006). Each algorithm was parameterized as follows: GLM and MARS, with formula = “quadratic” and interaction level = 3; GBM, with 10-fold cross validation procedure, maximum n° of trees = 5000, interaction depth = 3, bag fraction = 0.5 and shrinkage = 0.001; Maxent, with maximum n° of iterations = 2000, setting the remaining parameters as default.

Model evaluation and Ensemble Forecast For each of the models built on each of the 10 sets of presence-pseudoabsence data, 5 evaluation runs were performed, using 80% of the data to calibrate the model and the remaining 20% for the evaluation, resulting in a total number of 200 models for the current conditions (= 4 algorithms × 10 sets of presence-pseudoabsence data × 5 evaluation runs). Default prevalence = 0.5 in the “BIOMOD_Modeling” function was set, so that the weighted sum of the presences equals the weighted sum of the absences for each model.

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 432 Invasive common slider and global warming threaten European pond turtle

Model discrimination was evaluated through both the Area Under the Curve (AUC) of the Receiver Operating Characteristic Curve (ROC) (Phillips et al. 2006) and the True Skill Statistic (TSS) (Allouche et al. 2006). The ensemble modelling was performed for all the initial models which achieved an AUC value > 0.8 and a TSS value > 0.7. The algorithms used to build the ensemble models (EMs) were: “weighted mean of probabilities” (wmean), which results in EMs whose predictions are obtained by averaging the predictions of the single models after assigning them weights based on their AUC and TSS values (Thuiller et al. 2016); “coefficient of variation of probabilities” (cv), which permits to estimate the uncertainty within the predictions of the EM; “median of probabilities”, which is reported to be less sensitive to outliers with respect to the mean (Thuiller et al. 2016). Model extrapolation degree within the different future scenarios was checked through the “mess” function of the dismo R package (Elith et al. 2010; Hijmans et al. 2017). Subsequently, the raster maps resulting from the projection of the obtained EMs to the future scenarios predicted by the four different GCMs were processed through the MEDI algorithm (Iannella et al. 2017), a weighted-averaging procedure used to deal with differences in model extrapolation requirements to each future scenario. Moreover, given that changes in the predictors’ correlation structure between projecting and calibrating datasets may add uncertainties to SDMs’ projections (Mesgaran et al. 2014), we built correlation matrices to easily visualize noticeable changes in the predictors’ correlations. The algorithm used to build such matrices is reported, along with the matrices themselves, in Appendix 2. All raster calculations and corresponding cartographic analyses were performed in ArcMap 10.0. Variable importance within the EMs was assessed through the algorithm-independent randomization procedure implemented in biomod2 (Thuiller et al. 2009; Bucklin et al. 2015). The number of permutations to be performed in order to estimate the contribution of each predictor was set to 5. The raw mean contributions resulting from the 5 randomization runs were normalized by dividing the importance score of each predictor by the sum of the importance scores of all the 11 predictors selected for modelling (Bucklin et al. 2015), and then by multiplying the resulting quotient by 100, to obtain percent contribution scores. Loss, gain or stability of the areas predicted as suitable for the two species over time (current vs 2050 and 2070, for each of the considered RCPs) were measured by means of the “BIOMOD_RangeSize” function. Since this procedure needs binarized predictions (i.e., suitable vs unsuitable areas), we selected a binarization threshold for each species through the ecospat R package (Di Cola et al. 2017), maximising the TSS on the EM predictions. In fact, previous research demonstrated that maximization of the TSS in presence-background models permits selection of the same

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 433 Invasive common slider and global warming threaten European pond turtle

threshold as would be selected with presence-absence data, resulting in a more reliable binarization (Liu et al. 2013). Binary raster maps were further processed in ArcMap 10.0, and the extents of the three range shift categories (Loss, Stable, Gain) within the different future scenarios were calculated. Gap analysis (Scott et al. 1993; Jennings 2000) was performed in ArcMap 10.0 to investigate how the predicted changes of suitability patterns within the study region for E. orbicularis and T. scripta, under the different climate change scenarios considered, could affect the Italian network of protected areas (PAs). The shapefiles of the PAs were downloaded from the geo-portal of the Italian Ministry of the Environment (http://www.pcn.minambiente.it). Finally, the shapefiles resulting from the estimates of range size changes were further analysed to highlight possible shifts of E. orbicularis and T. scripta along the altitude gradient, using a 1 km2-resolution Digital Elevation Model (DEM), downloaded from the geo-portal of the Italian Ministry of the Environment (http://www.pcn.minambiente.it) and clipped to the boundaries of each category (i.e. Loss, Stable, Gain) in ArcMap 10.0. This procedure was performed to estimate the magnitude of variations in the suitable area (lost, maintained or gained) for the two species at increasing altitude intervals.

Niche comparison We used the “ecospat.niche.overlap”, “ecospat.niche.equivalency.test” and “ecospat.niche.similarity.test” functions from the ecospat package (Di Cola et al. 2017) to test the level of similarity between the environmental niches (Warren et al. 2008; McCormack et al. 2010) of E. orbicularis and T. scripta. The observed value of niche overlap between the EMs obtained for the two species under current conditions, expressed by the Schoener’s D metric (Schoener 1968), was compared to the 2.5th and 97.5th quantile of the null distributions obtained from the niche equivalency and similarity tests (1000 replicates) (Warren et al. 2008). In the first test, the null hypothesis is that observed niche overlap is greater than niche overlap between the simulated niches, while the null hypothesis of the niche similarity test is that the observed niche overlap is greater than the overlap between the niche of the native species and the simulated niches from the background of the invasive species (Di Cola et al. 2017). The selection of background points needed for each species in the background test was performed through the “Sampling Design Tool” in ArcMap 10.0 (Buja and Menza 2013).

Results The proximity analyses resulted in 30 co-occurrences (9.6% of the total 314 records for E. orbicularis) when considering a 1 km radius, and 93 co-occurrences when considering a 10 km radius (29.6%). The number of

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 434 Invasive common slider and global warming threaten European pond turtle

presence records after the spatial thinning was 272 for E. orbicularis and 183 for T. scripta, respectively the 86.6% and the 72.9% of the initial dataset available for each species falling within the study region (Figure 1). Further, the spatial thinning of native range dataset for T. scripta resulted in 1002 presence localities out of the initial 3757; a total of 1185 records were finally used for T. scripta model calibration. Checking for spatial autocorrelation within the thinned presence data sets of the study area, we obtained Moran’s I = 0.027 (expected value = −0.004), with z-score = 1.269 and P = 0.204, for E. orbicularis, and Moran’s I = 0.026 (expected value = −0.006), with z-score = 0.792 and P = 0.428 for T. scripta. Eleven bioclimatic variables (BIO2, BIO4, BIO5, BIO6, BIO8, BIO9, BIO10, BIO12, BIO16, BIO17 and BIO19) were selected as predictors after having checked for multicollinearity. Correlation matrices (Appendix 2) show differences among the considered GCMs in correlation changes between projection and calibration conditions of predictors; indeed, for CCSM4 and BCC-CSM-1 some pairs of predictors revealed inversion of the correlation trends, while the correlation structure remained more consistent for CNRM-CM5 and MIROC. However, none of the highly contributing variables for both species (see below) seem to be strongly affected by correlation changes; these patterns, together with the use of the MEDI algorithm to deal with extrapolation issues, reduced the uncertainties related to future predictions. The wmean maps show that a portion of the respective predicted suitable areas overlap for the two species (Figure 2A); the ratio between the overlap area and the area predicted as suitable for each species, computed considering the binarized maps obtained after applying the selected thresholds (see below), was 19.6% for E. orbicularis and 90.1% for T. scripta. The low values of cv (≤ 0.1) throughout the maps for both E. orbicularis and T. scripta confirm the low degree of uncertainty of the suitability patterns predicted by the EMs (Figure 2A). Furthermore, the high similarity between the median maps (Figure S3) and the wmean maps strengthens the reliability of the EMs’ predictions. The output of the niche equivalency test allows us to reject the null hypothesis that the environmental niches of E. orbicularis and T. scripta are indistinguishable: in fact, the observed niche overlap value was D = 0.307, well outside the 97.5th percentile of the null distribution

(D2,5% = 0.204, D97.5% = 0.264, p = 0.001). The results of the niche similarity test, however, did not confirm that the niches of two species are more similar than expected based on their habitat availability, considering that the observed D value falls between the 2.5th and the 97.5th percentile of the

null distribution obtained (D2,5% = 0.009, D97.5% = 0.498, p = 0.004). Results of both the niche tests are also graphically illustrated through the histograms shown in Figure 2B, 2C. The four highest normalized variable importance scores resulting from the wmean EMs for the two species are: for E. orbicularis, BIO10 = 39.4%,

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 435 Invasive common slider and global warming threaten European pond turtle

Figure 2. A) Weighted mean (larger) and coefficient of variation (smaller) maps resulting from the ensemble models obtained for the two target species; B) histogram of Schoener’ D null distribution resulting from niche equivalency test performed on the two target species; C) histogram of Schoener’ D null distribution resulting from niche similarity test. For both B) and C), the black arrow represents the observed niche overlap value, while the dashed lines define the 2.5th and the 97.5th percentile of the obtained null distributions.

BIO5 = 12.2%, BIO9 = 11.3% and BIO19 = 10.5%; for T. scripta, BIO10 = 42.8%, BIO16 = 20.5%, BIO5 = 10.4% and BIO2 = 8.0%. The corresponding response curves are shown in Figure S2. The two species share BIO10 as the highest contributing variable, with a similar trend, even though the positive relationship with predicted suitability is more accentuated for T. scripta. Moreover, E. orbicularis and T. scripta share another temperature-linked variable (BIO5), among the highest contributing ones, notwithstanding some differences in the respective trends. The binarization thresholds selected through the TSS-maximization procedure were 0.550 for E. orbicularis and 0.490 for T. scripta; we chose a single threshold (= 0.540) for which the TSS was close to the optimal one (difference lower than 0.1) for both species, to make the measures of range

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 436 Invasive common slider and global warming threaten European pond turtle

shifts less dependent upon the peculiarities of each species’ EM and to facilitate comparisons. The maps representing the range shifts predicted from the projection of the wmean EMs to the different future scenarios are shown in Figure 3, accompanied by histograms quantifying the extent of each range shift category (i.e. Loss, Stable, Gain). While for E. orbicularis a noticeable loss of suitable habitat is predicted for both the 2050 and 2070 RCP8.5 scenarios, T. scripta is predicted to greatly extend its suitable area under all the scenarios. Particularly, Figure 4A shows, for both the species and for each future scenario, the extent of the area predicted as stable in PAs, as well as the difference between gained and lost areas within PAs. Figure 4B shows the ratio between the overlapping suitable areas for the two target species and E. orbicularis’ suitable area (light grey). Moreover, it reports the ratio between the aforementioned overlapping areas falling within PAs and the corresponding PAs’ extent (grey). The ratio between T. scripta suitable area and the extent of PAs hosting these suitable conditions is also shown (dark grey). Both species are predicted to gain suitable areas within PAs under all the considered future scenarios, with T. scripta showing a clear increase along the RCP gradient (Figure 4A). First, under each future scenario, more than the 50% of the extent of suitable areas for E. orbicularis correspond to areas predicted as suitable also for T. scripta (Figure 4B). Second, areas predicted as suitable for both species within PAs under the 2050 scenarios cover more than 20% of the total extent of the PAs containing these overlapping suitable areas, and this percentage rises to more than 30% under the 2070_RCP4.5 scenario. Third, the percent extent of areas predicted as suitable for T. scripta within PAs with respect to the total extent of the PAs hosting suitable conditions for this species is higher than 30% for all the future scenarios except the 2050_RCP4.5 scenario. The altitude shift analysis for E. orbicularis (Figure 5A, C) for 2050 and 2070 indicates that this species would experience the loss of currently suitable areas mainly in flat and hilly territories (up to 600 metres), while the gain of suitable areas would occur mainly within the montane zone, with a further shift towards the subalpine zone (up to 1600 metres in 2050 and up to 1700 in 2070) under the RCP8.5 scenario. Analogously, Figure 5B, 5D highlight the predicted shift of suitable areas for T. scripta towards the montane regions for all the future scenarios considered.

Discussion One of the major threats to the survival of E. orbicularis is represented by T. scripta. It has been estimated that, prior to 1997, more than 900,000 individuals were imported in Italy (Ferri and Di Cerbo 2000), leading to a high propagule pressure on wide portions of the Italian territory.

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 437 Invasive common slider and global warming threaten European pond turtle

Figure 3. Maps reporting range shifts (red = area lost from the current to the projected scenario, yellow = stable area from current to the projected scenario, and green = area gained from current to the projected scenario) obtained from the four different ensemble projections (Relative Concentration Pathways (RCPs) to future scenarios (A: 2050_RCP4.5; B: 2050_RCP8.5; C: 2070_RCP4.5; D: 2070_RCP8.5)) for the two target species (1: E. orbicularis and 2: T. scripta). The histograms beside each map show the extent of the three range shift categories (in km2 × 1000).

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 438 Invasive common slider and global warming threaten European pond turtle

Figure 4. A) Area resulting from the gap analysis performed considering the Ensemble Models’ projections to future scenarios and the protected areas (PAs) of the Italian peninsula and Sardinia. Extent of stable areas and net gain (calculated as: areas gained – areas lost by the species under the future scenario) are depicted for each species. B) Percent extent of areas predicted as suitable for both E. orbicularis and T. scripta, with respect to the overall extent of areas predicted as suitable for E. orbicularis (light grey); percent extent of areas predicted as suitable for both species within PAs, with respect to the sum of the extents of the PAs containing these overlapping suitable areas (grey); percent extent of areas predicted as suitable for T. scripta within PAs, with respect to the sum of the extents of the PAs containing suitable areas for this species (dark grey).

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 439 Invasive common slider and global warming threaten European pond turtle

Figure 5. Extent of currently suitable area (black dash-dot curve) and extent of gained area (dash-dot curve), stable (continuous curve) and lost (dashed curve) suitable area under the RCP4.5 (blue curves) and RCP8.5 (red curves) emission scenarios along the altitudinal gradient, as resulting from the binarized ensemble models obtained under the current climatic conditions and their projections to the near future, for A) E. orbicularis in 2050, B) T. scripta in 2050, C) E. orbicularis in 2070 and D) T. scripta in 2070. The histogram bars for each of the three classes of range shift (gain, stable and loss), from which the B-spline curves were obtained, are also shown on the plot for each altitude break (200 meters).

Even though the proximity analyses showed that the number of present co-occurrences is relatively low within 1 km radius, the overlap analysis on the binarized habitat suitability revealed that almost the entire T. scripta suitable area falls within ranges suitable for E. orbicularis. On the flip side, a non-negligible portion of E. orbicularis suitable area is potentially intersected by T. scripta. From the projection of the EMs to the future scenarios, E. orbicularis is predicted to lose a large portion of currently suitable area, especially under the scenarios of high radiative forcing (RCP8.5). Gains observed under both the RCP4.5 future scenarios could counterbalance area lost by the species, while, for the RCP8.5, losses equal or overwhelm gains, with the depletion of wide continuous suitable territories in northern Italy. Gain of suitable territories predicted only for E. orbicularis is limited to some areas in the southern Piedmont and Liguria, southern Abruzzi, Molise and western side of Calabria, offering some potential refuge areas. However, these territories are not currently inhabited, or only to a limited extent, by this native pond turtle.

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 440 Invasive common slider and global warming threaten European pond turtle

In contrast, the potential invasiveness of T. scripta appears more extensive under all future scenarios, notwithstanding the conservative threshold value used to binarize the EM projections when investigating the possible future range shift of this species. Indeed, the substitution of the native E. orbicularis with the invasive T. scripta is predicted across some large areas (e.g. the Po valley) especially under the RCP8.5 scenario (see Figure 3B, 3D); in these areas, much of the T. scripta predicted gain will not directly affect E. orbicularis, which is predicted to concurrently disappear from them due to new unsuitable climatic conditions. The potentially suitable areas for T. scripta are predicted to continuously extend within all the considered scenarios, with very high rates of colonization (gain) and no predicted losses: the absence of losses could be due to the limited extent of suitable areas under current climate (Figure 2B), which remain stable in all the projections to future scenarios. These patterns are probably linked to the increase in maximum temperatures, which our EMs show to be highly influential for both E. orbicularis and T. scripta (BIO10 and BIO5). Our ensemble forecasts are in line with the predictions reported in Ficetola et al. (2009); their models find a negative influence of low precipitation values on predicted suitability, which is corroborated by the T. scripta response curve to precipitation of the wettest quarter (BIO16) (Figure S2). Low humidity (and the extreme case of droughts) has detrimental effects on T. scripta, especially for embryo development, hatchling success (Tucker and Paukstis 2000; Delmas et al. 2008), and adult resources availability (Friggens et al. 2013). Obviously, the existence of wetlands depends on a certain amount of precipitation. The EMs obtained for T. scripta confirm the importance of BIO16 and BIO5 in defining suitable conditions for this species; as reported in Ficetola et al. (2009), the temperature of the warmest month (and reasonably the temperature of the warmest quarter [BIO10]) plays a key role in both reproductive and non-reproductive populations, and precipitations are positively linked with the occurrence of T. scripta. Furthermore, Ficetola et al. (2009) forecasts about the probable expansion of this IAS associated with global climate change are confirmed by our projections and extended to a longer time span. The overlapping suitable area to both the species under both the current and the future scenarios (cf. Figure 2, 4B), together with the confirmation of a slight similarity between the environmental niches of E. orbicularis and T. scripta (D = 0.307), even though not confirmed as independent from the respective backgrounds (see Results), represent important warnings for the conservation of E. orbicularis (Vicente et al. 2011). In addition, a partial overlap of trophic niches between the two species was recently observed in Italy by Balzani et al. (2016), further jeopardizing E. orbicularis’ invaded populations. The role of PAs in the future will be of crucial importance, especially when they will have to deal with habitat or species management and spatial

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 441 Invasive common slider and global warming threaten European pond turtle

planning of their territories in response to IAS and climate change (Araújo et al. 2011). In both 2050 and 2070, an increase of suitable areas for T. scripta is expected to be observed under both RCP4.5 and RCP8.5, corresponding to a potential sympatry in PAs with E. orbicularis (cf. Figure 4B), for which the predicted suitable areas within PAs tend to increase in a similar way (cf. Figure 4A). The gain counterbalancing the loss of suitable area for E. orbicularis in all the future scenarios, except the 2070_8.5 one, may seem positive at first, but it must be interpreted with some considerations about which areas are predicted to become suitable. In fact, the range shift of E. orbicularis is accompanied by an altitude shift, a phenomenon observed for many other taxa (Gottfried et al. 1999; Chen et al. 2011; Stanisci et al. 2016; Scherrer et al. 2017; Urbani et al. 2017); this potential upward displacement is predicted also for T. scripta (Figure 5). In this context, only some isolated patches seem to offer refuge areas to E. orbicularis, which is predicted to gain new territories at higher altitudes than T. scripta in the southern part of the Apennines, while the altitudinal shifts of the two species in many other territories would lead to potential co-occurrence. This means that the two species would potentially have to compete in many new areas, a challenge in which T. scripta will be probably favoured because that species is not habitat specific (Gibbons 1990) and shows high dispersal capabilities (Ryan et al. 2008).

Conclusions The future scenarios emerging from our analyses are of concern for the conservation of the native E. orbicularis. Both the predicted current and future suitable areas for the European pond turtle are analogously potentially suitable for its strong competitor T. scripta, in a large portion of the Italian peninsula and Sardinia (i.e. more than half of the total predicted suitable areas for E. orbicularis). As highlighted through the gap analysis, PAs may soon have to oversee specific strategies to deal with the potential sympatry of E. orbicularis and T. scripta. Our projections show that many mountainous areas, which are currently protected by nationally- or internationally-established PAs, may become suitable for both species, potentially threatening the remaining E. orbicularis’ populations. Indeed, while this species only migrates to new suitable areas through natural ecological corridors, T. scripta may gradually invade new areas via human-mediated releases, nullifying habitat management and native species protection. Nevertheless, predictions of future range shifts should not be accepted on blind faith, since it has been shown that correlative SDMs may fail in accurately predict range expansion or contraction under future climate scenarios (Sofaer et al. 2018). This should encourage future studies deepening the knowledge on

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 442 Invasive common slider and global warming threaten European pond turtle

the ecophysiological responses of E. orbicularis and T. scripta in a climate- change context, to more fairly evaluate SDMs’ predictions of potential future sympatry areas. Moreover, further research is needed to increase the amount of data regarding reproductive sites for T. scripta to more easily investigate biotic and abiotic factors influencing successful reproduction in invaded territories. We suggest that further fine-scale analyses will have to be conducted by authorities of PAs in the next years, followed by updated management plans to address specific conservation issues raised by the non-native T. scripta. Finally, we recommend extensive environmental education campaigns specifically addressed to the problems that IAS cause to native fauna, such as avoiding the release of animals in protected areas.

Acknowledgements

We are particularly thankful to the two anonymous reviewers and to all our collaborators who helped collect data during literature search and field efforts.

References

Aiello‐Lammens ME, Boria RA, Radosavljevic A, Vilela B, Anderson RP (2015) spThin: an R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography 38: 541‒545, https://doi.org/10.1111/ecog.01132 Allouche O, Tsoar A, Kadmon R (2006) Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). Journal of Applied Ecology 43: 1223‒1232, https://doi.org/10.1111/j.1365-2664.2006.01214.x Andreone F, Corti C, Ficetola F, Razzetti E, Romano A, Sindaco R (2013) Emys orbicularis. IUCN Red List Italia. http://www.webcitation.org/6rO6CpSQ4 (accessed: 2017‒06‒21) Araújo MB, Alagador D, Cabeza M, Nogués‐Bravo D, Thuiller W (2011) Climate change threatens European conservation areas. Ecology Letters 14: 484‒492, https://doi.org/10.1111/ j.1461-0248.2011.01610.x Ashraf U, Peterson AT, Chaudhry MN, Ashraf I, Saqib Z, Ahmad SR, Ali H (2017) Ecological niche model comparison under different climate scenarios: a case study of Olea spp. in Asia. Ecosphere 8: e01825, https://doi.org/10.1002/ecs2.1825 Balzani P, Vizzini S, Santini G, Masoni A, Ciofi C, Ricevuto E, Chelazzi G (2016) Stable isotope analysis of trophic niche in two co-occurring native and invasive terrapins, Emys orbicularis and Trachemys scripta elegans. Biological Invasions 18: 3611‒3621, https://doi.org/10.1007/s10530-016-1251-x Bellard C, Thuiller W, Leroy B, Genovesi P, Bakkenes M, Courchamp F (2013) Will climate change promote future invasions? Global Change Biology 19: 3740‒3748, https://doi.org/ 10.1111/gcb.12344 Berg JA, Meyer GA, Young EB (2016) Propagule pressure and environmental conditions interact to determine establishment success of an invasive plant species, glossy buckthorn (Frangula alnus), across five different wetland habitat types. Biological Invasions 18: 1363‒1373, https://doi.org/10.1007/s10530-016-1073-x Bomford M, Barry SC, Lawrence E (2010) Predicting establishment success for introduced freshwater fishes: a role for climate matching. Biological Invasions 12: 2559‒2571, https://doi.org/10.1007/s10530-009-9665-3 Brandt LA, Benscoter AM, Harvey R, Speroterra C, Bucklin D, Romañach SS, Watling JI, Mazzotti FJ (2017) Comparison of climate envelope models developed using expert- selected variables versus statistical selection. Ecological Modelling 345: 10‒20, https://doi.org/10.1016/j.ecolmodel.2016.11.016 Broennimann O, Guisan A (2008) Predicting current and future biological invasions: both native and invaded ranges matter. Biology Letters 4: 585‒589, https://doi.org/10.1098/rsbl. 2008.0254 Bucklin DN, Basille M, Benscoter AM, Brandt LA, Mazzotti FJ, Romanach SS, Speroterra C, Watling JI (2015) Comparing species distribution models constructed with different subsets of environmental predictors. Diversity and Distributions 21: 23‒35, https://doi.org/10.1111/ ddi.12247 Buisson L, Thuiller W, Casajus N, Lek S, Grenouillet G (2010) Uncertainty in ensemble forecasting of species distribution. Global Change Biology 16: 1145‒1157, https://doi.org/10. 1111/j.1365-2486.2009.02000.x

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 443 Invasive common slider and global warming threaten European pond turtle

Buja K, Menza C (2013) Sampling design tool for ArcGIS: Instruction manual. [for ESRI ArcGIS 10.0 Service Pack 3 or higher]. NOAA/National Centers for Coastal Ocean Science, Silver Spring, MD. http://aquaticcommons.org/14676/1/Buja%20and%20Menza%202013.pdf Burke VJ, Greene JL, Gibbons JW (1995) The effect of sample size and study duration on metapopulation estimates for slider turtles (Trachemys scripta). Herpetologica 51: 451‒456 Cadi A, Joly P (2003) Competition for basking places between the endangered European pond turtle (Emys orbicularis galloitalica) and the introduced red-eared slider (Trachemys scripta elegans). Canadian Journal of Zoology 81: 1392‒1398, https://doi.org/10.1139/z03-108 Cadi A, Joly P (2004) Impact of the introduction of the red-eared slider (Trachemys scripta elegans) on survival rates of the European pond turtle (Emys orbicularis). Biodiversity and Conservation 13: 2511‒2518, https://doi.org/10.1023/b:bioc.0000048451.07820.9c Caffrey JM, Baars J-R, Barbour JH, Boets P, Boon P, Davenport K, Dick JT, Early J, Edsman L, Gallagher C, Gross J, Heinimaa P, Horrill C, Hudin S, Hulme P, Hynes S, MacIsaac H, McLoone P, Millane M, Moen T, Moore N, Newman J, O’Conchuir R, O’Farrell M, O’Flynn C, Oidtmann B, Renals T, Ricciardi A, Roy H, Shaw R, Weyl O, Williams F, Lucy F (2014) Tackling invasive alien species in Europe: the top 20 issues. Management of Biological Invasions 5: 1‒20, https://doi.org/10.3391/mbi.2014.5.1.01 Cerasoli F, Iannella M, D’Alessandro P, Biondi M (2017) Comparing pseudo-absences generation techniques in Boosted Regression Trees models for conservation purposes: A case study on amphibians in a protected area. PLoS ONE 12: e0187589, https://doi.org/10.1371/journal.pone.0187589 Chen I-C, Hill JK, Ohlemüller R, Roy DB, Thomas CD (2011) Rapid range shifts of species associated with high levels of climate warming. Science 333: 1024‒1026, https://doi.org/10.1126/science.1206432 Crase B, Liedloff AC, Wintle BA (2012) A new method for dealing with residual spatial autocorrelation in species distribution models. Ecography 35: 879‒888, https://doi.org/10. 1111/j.1600-0587.2011.07138.x Delmas V, Bonnet X, Girondot M, Prévot-Julliard A-C (2008) Varying hydric conditions during incubation influence egg water exchange and hatchling phenotype in the red-eared slider turtle. Physiological and Biochemical Zoology 81: 345‒355, https://doi.org/10.1086/ 529459 Di Cola V, Broennimann O, Petitpierre B, Breiner FT, D’Amen M, Randin C, Engler R, Pottier J, Pio D, Dubuis A (2017) ecospat: An R package to support spatial analyses and modeling of species niches and distributions. Ecography 40: 774‒787, https://doi.org/10.1111/ecog.02671 Di Tizio L, Di Cerbo AR (2011) Trachemys scripta. In: Corti C, Capula M, Luiselli L, Razzetti E, Sindaco R (eds), Fauna d’Italia - Reptilia. Calderini Bologna, pp 170‒179 Dormann CF (2007) Effects of incorporating spatial autocorrelation into the analysis of species distribution data. Global Ecology and Biogeography 16: 129‒138, https://doi.org/10.1111/ j.1466-8238.2006.00279.x Dormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carré G, Marquéz JRG, Gruber B, Lafourcade B, Leitão PJ (2013) Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36: 27‒46, https://doi.org/10.1111/ j.1600-0587.2012.07348.x EEA (2012) The impacts of invasive alien species in Europe. European Environment Agency, Technical Report, Copenhagen, Denmark, 114 pp, https://doi.org/10.2800/65864 Elith J, Graham CH, Anderson RP, Dudik M, Ferrier S, Guisan A, Hijmans RJ, Huettman F, Leathwick JR, Lehman A, Jin L, Lohman LG, Loiselle BA, Peterson AT, Phillips SJ (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29: 129‒151, https://doi.org/10.1111/j.2006.0906-7590.04596.x Elith J, Leathwick JR, Hastie T (2008) A working guide to boosted regression trees. Journal of Animal Ecology 77: 802‒813, https://doi.org/10.1111/j.1365-2656.2008.01390.x Elith J, Kearney M, Phillips S (2010) The art of modelling range‐shifting species. Methods in Ecology and Evolution 1: 330‒342, https://doi.org/10.1111/j.2041-210x.2010.00036.x ESRI (2010) ArcMap 10.0. ESRI, Redlands, California European Union (1997) Wildlife Trade Regulation, EC/338/97 Ferreira AF, Quintella BR, Maia C, Mateus CS, Alexandre CM, Capinha C, Almeida PR (2013) Influence of macrohabitat preferences on the distribution of European brook and river lampreys: Implications for conservation and management. Biological Conservation 159: 175‒186, https://doi.org/10.1016/j.biocon.2012.11.013 Ferri V, Di Cerbo A (2000) La Trachemys scripta elegans (Wied, 1839) negli ambienti umidi lombardi: inquinamento faunistico o problema ecologico. Atti 1 Congresso nazionale Societas Herpetologica Italica, Torino, pp 2‒6, https://doi.org/10.1002/mmnz.20020780114 Ficetola GF, Thuiller W, Miaud C (2007) Prediction and validation of the potential global distribution of a problematic alien invasive species - the American bullfrog. Diversity and Distributions 13: 476‒485, https://doi.org/10.1111/j.1472-4642.2007.00377.x Ficetola GF, Thuiller W, Padoa-Schioppa E (2009) From introduction to the establishment of alien species: bioclimatic differences between presence and reproduction localities in the slider turtle. Diversity and Distributions 15: 108‒116, https://doi.org/10.1111/j.1472-4642.2008. 00516.x

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 444 Invasive common slider and global warming threaten European pond turtle

Fourcade Y, Engler JO, Rodder D, Secondi J (2014) Mapping species distributions with MAXENT using a geographically biased sample of presence data: a performance assessment of methods for correcting sampling bias. PLoS ONE 9: e97122, https://doi.org/ 10.1371/journal.pone.0097122 Friggens MM, Finch DM, Bagne KE, Coe SJ, Hawksworth DL (2013) Vulnerability of species to climate change in the Southwest: terrestrial species of the Middle Rio Grande. In: U.S. Department of Agriculture Fs, Rocky Mountain Research station (ed), USDA, pp 191, https://doi.org/10.2737/rmrs-gtr-306 Garcia RA, Burgess ND, Cabeza M, Rahbek C, Araujo MB (2012) Exploring consensus in 21st century projections of climatically suitable areas for African vertebrates. Global Change Biology 18: 1253‒1269, https://doi.org/10.1111/j.1365-2486.2011.02605.x Gent PR, Danabasoglu G, Donner LJ, Holland MM, Hunke EC, Jayne SR, Worley PH (2011) The community climate system model version 4. Journal of Climate 24: 4973‒4991, https://doi.org/10.1175/2011jcli4083.1 Gibbons JW (1990) Life history and ecology of the slider turtle. Smithsonian Institution Press Washington, DC, 367 pp Gilioli G, Schrader G, Carlsson N, van Donk E, van Leeuwen CH, Martín PR, Pasquali S, Vilà M, Vos S (2017) Environmental risk assessment for invasive alien species: A case study of apple snails affecting ecosystem services in Europe. Environmental Impact Assessment Review 65: 1‒11, https://doi.org/10.1016/j.eiar.2017.03.008 Gottfried M, Pauli H, Reiter K, Grabherr G (1999) A fine‐scaled predictive model for changes in species distribution patterns of high mountain plants induced by climate warming. Diversity and Distributions 5: 241‒251, https://doi.org/10.1046/j.1472-4642.1999.00058.x Grenouillet G, Buisson L, Casajus N, Lek S (2011) Ensemble modelling of species distribution: the effects of geographical and environmental ranges. Ecography 34: 9‒17, https://doi.org/ 10.1111/j.1600-0587.2010.06152.x Guillera-Arroita G, Lahoz-Monfort JJ, Elith J, Gordon A, Kujala H, Lentini PE, McCarthy MA, Tingley R, Wintle BA (2015) Is my species distribution model fit for purpose? Matching data and models to applications. Global Ecology and Biogeography 24: 276‒292, https://doi.org/10.1111/geb.12268 Guisan A, Tingley R, Baumgartner JB, Naujokaitis-Lewis I, Sutcliffe PR, Tulloch AI, Regan TJ, Brotons L, McDonald-Madden E, Mantyka-Pringle C, Martin TG, Rhodes JR, Maggini R, Setterfield SA, Elith J, Schwartz MW, Wintle BA, Broennimann O, Austin M, Ferrier S, Kearney MR, Possingham HP, Buckley YM (2013) Predicting species distributions for conservation decisions. Ecology Letters 16: 1424‒35, https://doi.org/10.1111/ele.12189 Hidalgo-Vila J, Díaz-Paniagua C, Ribas A, Florencio M, Pérez-Santigosa N, Casanova J (2009) Helminth communities of the exotic introduced turtle, Trachemys scripta elegans in southwestern Spain: Transmission from native turtles. Research in Veterinary Science 86: 463‒465, https://doi.org/10.1016/j.rvsc.2008.08.003 Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965‒1978, https://doi.org/10.1002/joc.1276 Hijmans RJ, Phillips S, Leathwick J , Elith J (2017) dismo: Species Distribution Modeling. R package version 1.1-4 Iannella M, Cerasoli F, Biondi M (2017) Unraveling climate influences on the distribution of the parapatric newts Lissotriton vulgaris meridionalis and L. italicus. Frontiers in Zoology 14: 55, https://doi.org/10.1186/s12983-017-0239-4 Iannella M, Cerasoli F, D’Alessandro P, Console G, Biondi M (2018a) Coupling GIS spatial analysis and Ensemble Niche Modelling to investigate climate change-related threats to the Sicilian pond turtle Emys trinacris, an endangered species from the Mediterranean. PeerJ 6: e4969, https://doi.org/10.7717/peerj.4969 Iannella M, D’Alessandro P, Biondi M (2018b) Evidences for a shared history for spectacled salamanders, haplotypes and climate. Scientific Reports 8: 16507, https://doi.org/10.1038/s41598- 018-34854-1 Jennings MD (2000) Gap analysis: concepts, methods, and recent results. Landscape Ecology 15: 5‒20 Khanum R, Mumtaz AS, Kumar S (2013) Predicting impacts of climate change on medicinal asclepiads of Pakistan using Maxent modeling. Acta Oecologica-International Journal of Ecology 49: 23‒31, https://doi.org/10.1016/j.actao.2013.02.007 Kraus F (2009) Alien reptiles and amphibians: a scientific compendium and analysis. Springer Science & Business Media, 563 pp Leathwick JR, Rowe D, Richardson J, Elith J, Hastie T (2005) Using multivariate adaptive regression splines to predict the distributions of New Zealand’s freshwater diadromous fish. Freshwater Biology 50: 2034‒2052, https://doi.org/10.1111/j.1365-2427.2005.01448.x Liu CR, White M, Newell G (2013) Selecting thresholds for the prediction of species occurrence with presence-only data. Journal of Biogeography 40: 778‒789, https://doi.org/ 10.1111/jbi.12058

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 445 Invasive common slider and global warming threaten European pond turtle

Loiselle BA, Howell CA, Graham CH, Goerck JM, Brooks T, Smith KJ, Williams PH (2003) Avoiding Pitfalls of Using Species Distribution Models in Conservation Planning. Conservation Biology 17: 1591‒1600, https://doi.org/10.1111/j.1523-1739.2003.00233.x McCormack JE, Zellmer AJ, Knowles LL (2010) Does niche divergence accompany allopatric divergence in Aphelocoma jays as predicted under ecological speciation?: insights from tests with niche models. Evolution 64: 1231‒1244, https://doi.org/10.1111/j.1558-5646.2009. 00900.x McCullagh P (1984) Generalized linear models. European Journal of Operational Research 16: 285‒292 Meehl GA, Goddard L, Murphy J, Stouffer RJ, Boer G, Danabasoglu G, Dixon K, Giorgetta MA, Greene AM, Hawkins E, Hegerl G, Karoly D, Keenlyside N, Kimoto M, Kirtman B, Navarra A, Pulwarty R, Smith D, Stammer D, Stockdale T (2009) Decadal prediction: can it be skillful? Bulletin of the American Meteorological Society 90: 1467‒1485, https://doi.org/ 10.1175/2009bams2778.1 Meinshausen M, Smith SJ, Calvin K, Daniel JS, Kainuma M, Lamarque J, Matsumoto K, Montzka S, Raper S, Riahi K (2011) The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change 109: 213, https://doi.org/10.1007/s10584-011- 0156-z Mesgaran MB, Cousens RD, Webber BL (2014) Here be dragons: a tool for quantifying novelty due to covariate range and correlation change when projecting species distribution models. Diversity and Distributions 20: 1147–1159, https://doi.org/10.1111/ddi.12209 Meyer L, Du Preez L, Verneau O, Bonneau E, Héritier L (2015) Parasite host-switching from the invasive American red-eared slider, Trachemys scripta elegans, to the native Mediterranean pond turtle, Mauremys leprosa, in natural environments. Aquatic Invasions 10: 79‒91, https://doi.org/10.3391/ai.2015.10.1.08 Naimi B, Araújo MB (2016) sdm: a reproducible and extensible R platform for species distribution modelling. Ecography 39: 368‒375, https://doi.org/10.1111/ecog.01881 Occhipinti-Ambrogi A (2007) Global change and marine communities: alien species and climate change. Marine Pollution Bulletin 55: 342‒352, https://doi.org/10.1016/j.marpolbul. 2006.11.014 Oi M, Araki J, Matsumoto J, Nogami S (2012) Helminth fauna of a turtle species introduced in Japan, the red-eared slider turtle (Trachemys scripta elegans). Research in Veterinary Science 93: 826‒830, https://doi.org/10.1016/j.rvsc.2011.10.001 Oreska MP, Aldridge DC (2011) Estimating the financial costs of freshwater invasive species in Great Britain: a standardized approach to invasive species costing. Biological Invasions 13: 305‒319, https://doi.org/10.1007/s10530-010-9807-7 Peterson AT (2003) Predicting the geography of species’ invasions via ecological niche modeling. The Guarterly Review of Biology 78: 419‒33, https://doi.org/10.1086/378926 Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modeling of species geographic distributions. Ecological Modelling 190: 231‒259, https://doi.org/10.1016/j.ecolmodel. 2005.03.026 Polo-Cavia N, López P, Martín J (2014) Interference competition between native Iberian turtles and the exotic Trachemys scripta. Basic and Applied Herpetology 28: 5‒20, https://doi.org/ 10.11160/bah.13014 Poulos HM, Chernoff B, Fuller PL, Butman D (2012) Ensemble forecasting of potential habitat for three invasive fishes. Aquatic Invasions 7: 59‒72, https://doi.org/10.3391/ai.2012.7.1.007 R Core Team (2016) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/ Riahi K, Rao S, Krey V, Cho CH, Chirkov V, Fischer G, Kindermann G, Nakicenovic N, Rafaj P (2011) RCP 8.5-A scenario of comparatively high greenhouse gas emissions. Climatic Change 109: 33‒57, https://doi.org/10.1007/s10584-011-0149-y Rissler LJ, Apodaca JJ (2007) Adding more ecology into species delimitation: ecological niche models and phylogeography help define cryptic species in the black salamander (Aneides flavipunctatus). Systematic Biology 56: 924‒942, https://doi.org/10.1080/10635150701703063 Rodder D, Schmidtlein S, Veith M, Lotters S (2009) Alien invasive slider turtle in unpredicted habitat: a matter of niche shift or of predictors studied? PLoS ONE 4: e7843, https://doi.org/ 10.1371/journal.pone.0007843 Ryan TJ, Conner CA, Douthitt BA, Sterrett SC, Salsbury CM (2008) Movement and habitat use of two aquatic turtles (Graptemys geographica and Trachemys scripta) in an urban landscape. Urban Ecosystems 11: 213‒225, https://doi.org/10.1007/s11252-008-0049-8 Scalera R (2009) Species account of 100 of the Most Invasive Alien Species in Europe. Handbook of Alien Species in Europe, Springer, Berlin, 374 pp, https://doi.org/10.1007/978-1- 4020-8280-1_13 Scherrer D, Massy S, Meier S, Vittoz P, Guisan A (2017) Assessing and predicting shifts in mountain forest composition across 25 years of climate change. Diversity and Distributions 23: 517‒528, https://doi.org/10.1111/ddi.12548 Schoener TW (1968) The Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49: 704‒726, https://doi.org/10.2307/1935534

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 446 Invasive common slider and global warming threaten European pond turtle

Scott JM, Davis F, Csuti B, Noss R, Butterfield B, Groves C, Ulliman J (1993) Gap analysis: a geographic approach to protection of biological diversity. Wildlife Monographs 123: 3‒41 Seebens H, Blackburn TM, Dyer EE, Genovesi P, Hulme PE, Jeschke JM, Pagad S, Pyšek P, Winter M, Arianoutsou M (2017) No saturation in the accumulation of alien species worldwide. Nature Communications 8: 14435, https://doi.org/10.1038/ncomms14435 Sofaer HR, Jarnevich CS, Flather CH (2018) Misleading prioritizations from modelling range shifts under climate change. Global Ecology and Biogeography 27: 658–666, https://doi.org/10.1111/geb.12726 Sorte CJ, Ibáñez I, Blumenthal DM, Molinari NA, Miller LP, Grosholz ED, Diez JM, D'Antonio CM, Olden JD, Jones SJ (2013) Poised to prosper? A cross‐system comparison of climate change effects on native and non‐native species performance. Ecology Letters 16: 261‒270, https://doi.org/10.1111/ele.12017 Stanisci A, Frate L, Morra Di Cella U, Pelino G, Petey M, Siniscalco C, Carranza M (2016) Short-term signals of climate change in Italian summit vegetation: observations at two GLORIA sites. Plant Biosystems-An International Journal Dealing with all Aspects of Plant Biology 150: 227‒235, https://doi.org/10.1080/11263504.2014.968232 Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (2014) Climate change 2013: The physical science basis. Cambridge University Press Cambridge, UK, and New York, 1523 pp Stohlgren TJ, Ma P, Kumar S, Rocca M, Morisette JT, Jarnevich CS, Benson N (2010) Ensemble habitat mapping of invasive plant species. Risk Analysis 30: 224‒235, https://doi.org/10.1111/j.1539-6924.2009.01343.x Stralberg D, Matsuoka SM, Hamann A, Bayne EM, Solymos P, Schmiegelow FK, Wang X, Cumming SG, Song SJ (2015) Projecting boreal bird responses to climate change: the signal exceeds the noise. Ecological Applications 25: 52‒69, https://doi.org/10.1890/13-2289.1 Telecky T (2001) United States import and export of live turtles and tortoises. Turtle and Tortoise Newsletter 4: 8‒13 TFTS Group (1996) Emys orbicularis. The IUCN Red List of Threatened Species 1996: e.T7717A97292665 (accessed 21 June 2017), https://doi.org/10.2305/iucn.uk.1996.rlts.t7717a128 44431.en Thuiller W, Araujo MB, Pearson RG, Whittaker RJ, Brotons L, Lavorel S (2004) Biodiversity conservation: uncertainty in predictions of extinction risk. Nature 430: 6995, https://doi.org/ 10.1038/nature02716 Thuiller W, Lafourcade B, Engler R, Araujo MB (2009) BIOMOD - a platform for ensemble forecasting of species distributions. Ecography 32: 369‒373, https://doi.org/10.1111/j.1600- 0587.2008.05742.x Thuiller W, Georges D, Engler R, Breiner F, Georges MD (2016) biomod2: Ensemble platform for species distribution modeling. R package version 3.3‒7 Treglia ML, Fisher RN, Fitzgerald LA (2015) Integrating Multiple Distribution Models to Guide Conservation Efforts of an Endangered Toad. PLoS ONE 10: e0131628, https://doi.org/10.1371/journal.pone.0131628 Tucker JK, Paukstis GL (2000) Hatching success of turtle eggs exposed to dry incubation environment. Journal of Herpetology 34: 529‒534, https://doi.org/10.2307/1565267 Urbani F, D’Alessandro P, Frasca R, Biondi M (2015) Maximum entropy modeling of geographic distributions of the flea beetle species endemic in Italy (Coleoptera: Chrysomelidae: Galerucinae: Alticini). Zoologischer Anzeiger - A Journal of Comparative Zoology 258: 99‒109, https://doi.org/10.1016/j.jcz.2015.08.002 Urbani F, D’Alessandro P, Biondi M (2017) Using Maximum Entropy Modeling (MaxEnt) to predict future trends in the distribution of high altitude endemic insects in response to climate change. Bulletin of Insectology 70: 189‒200 Urbina-Cardona JN, Flores-Villela O (2010) Ecological-niche modeling and prioritization of conservation-area networks for Mexican herpetofauna. Conservation Biology 24: 1031‒1041, https://doi.org/10.1111/j.1523-1739.2009.01432.x van Wilgen NJ, Richardson DM (2012) The roles of climate, phylogenetic relatedness, introduction effort, and reproductive traits in the establishment of non-native reptiles and amphibians. Conservation Biology 26: 267‒77, https://doi.org/10.1111/j.1523-1739.2011.01804.x Veloz SD (2009) Spatially autocorrelated sampling falsely inflates measures of accuracy for presence-only niche models. Journal of Biogeography 36: 2290‒2299, https://doi.org/10.1111/ j.1365-2699.2009.02174.x Vicente J, Randin CF, Gonçalves J, Metzger MJ, Lomba Â, Honrado J, Guisan A (2011) Where will conflicts between alien and rare species occur after climate and land-use change? A test with a novel combined modelling approach. Biological Invasions 13: 1209‒1227, https://doi.org/10.1007/s10530-011-9952-7 Voldoire A, Sanchez-Gomez E, Salas y Mélia D, Decharme B, Cassou C, Sénési S (2011) The CNRM-CM5.1 global climate model: description and basic evaluation. Climate Dynamics 40: 2091‒2121, https://doi.org/10.1007/s00382-011-1259-y Walther G-R, Roques A, Hulme PE, Sykes MT, Pyšek P, Kühn I, Zobel M, Bacher S, Botta- Dukát Z, Bugmann H (2009) Alien species in a warmer world: risks and opportunities. Trends in Ecology & Evolution 24: 686‒693, https://doi.org/10.1016/j.tree.2009.06.008

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 447 Invasive common slider and global warming threaten European pond turtle

Warren DL, Glor RE, Turelli M (2008) Environmental niche equivalency versus conservatism: quantitative approaches to niche evolution. Evolution 62: 2868‒2883, https://doi.org/10.1111/ j.1558-5646.2008.00482.x Watanabe S, Hajima T, Sudo K, Nagashima T, Takemura T, Okajima H, Nozawa T, Kawase H, Abe M, Yokohata T, Ise T, Sato H, Kato E, Takata K, Emori S, Kawamiya M (2011) MIROC-ESM 2010: model description and basic results of CMIP5-20c3m experiments. Geoscientific Model Development 4: 845‒872, https://doi.org/10.5194/gmd-4-845-2011 Werneck FP, Costa GC, Colli GR, Prado DE, Sites Jr JW (2011) Revisiting the historical distribution of Seasonally Dry Tropical Forests: new insights based on palaeodistribution modelling and palynological evidencegeb. Global Ecology and Biogeography 20: 272‒288, https://doi.org/10.1111/j.1466-8238.2010.00596.x Wu T, Song L, Li W, Wang Z, Zhang H, Xin X, Zhang Y, Zhang L, Li J, Wu F, Liu Y, Zhang F, Shi X, Chu M, Zhang J, Fang Y, Wang F, Lu Y, Liu X, Wei M, Liu Q, Zhou W, Dong M, Zhao Q, Ji J, Li L, Zhou M (2014) An overview of BCC climate system model development and application for climate change studies. Acta Meteorologica Sinica 28: 34‒56, https://doi.org/10.1007/s13351-014-3041-7 Zhang J, Nielsen SE, Chen YH, Georges D, Qin YC, Wang SS, Svenning JC, Thuiller W (2017) Extinction risk of North American seed plants elevated by climate and land-use change. Journal of Applied Ecology 54: 303‒312, https://doi.org/10.1111/1365-2664.12701 Zhang L, Liu S, Sun P, Wang T, Wang G, Zhang X, Wang L (2015) Consensus forecasting of species distributions: the effects of Niche model performance and Niche properties. PLoS ONE 10: e0120056, https://doi.org/10.1371/journal.pone.0120056 Zuffi MAL, Di Cerbo AR, Fritz U (2011) Emys orbicularis. In: Corti C, Capula M, Luiselli L, Razzetti E, Sindaco R (eds), Fauna d’Italia - Reptilia. Calderini Bologna, pp 153‒163

Supplementary material The following supplementary material is available for this article: Appendix 1. Codes and explication of the nineteen bioclimatic variables considered as candidate predictors (from Worldclim.org). Appendix 2. The algorithm used to explore changes within the predictors’ correlation structure between current climatic conditions and future scenarios predicted by the four considered Global Climate Models. Table S1. Species’ dataset used in SDMs calibration. Table S2. Correlation matrix used to select the predictors for the Ensemble Models. Values of autocorrelation exceeding the chosen threshold are highlighted (Pearson’s r > 0.85, in red, or r < −0.85, in yellow). Figure S1. Map of North America representing the native range of Trachemys scripta, with presence localities (for details, please see Table S1). The native range is based on the USGS’ map available at https://nas.er.usgs.gov (accessed February 19, 2018). Figure S2. Response curves of mean diurnal range (BIO2), maximum temperature of warmest month (BIO5), mean temperature of the driest quarter (BIO9), mean temperature of the warmest quarter (BIO10), precipitation of the wettest quarter (BIO16) and precipitation of the coldest quarter (BIO19), representing the variables with the highest normalized importance scores within the Ensemble Models obtained for A) Emys orbicularis and B) Trachemys scripta. Figure S3. Ensemble Models of median habitat suitability obtained for the Emys orbicularis and Trachemys scripta for the Italian peninsula and Sardinia. This material is available as part of online article from: http://www.reabic.net/journals/mbi/2019/Supplements/MBI_2019_Cerasoli_etal_Appendix1.xlsx http://www.reabic.net/journals/mbi/2019/Supplements/MBI_2019_Cerasoli_etal_Appendix2-Part1.pdf http://www.reabic.net/journals/mbi/2019/Supplements/MBI_2019_Cerasoli_etal_Appendix2-Part2.xlsx http://www.reabic.net/journals/mbi/2019/Supplements/MBI_2019_Cerasoli_etal_SupplementaryTables.xlsx http://www.reabic.net/journals/mbi/2019/Supplements/MBI_2019_Cerasoli_etal_SupplementaryFigures.pdf

Cerasoli et al. (2019), Management of Biological Invasions 10(3): 428–448, https://doi.org/10.3391/mbi.2019.10.3.02 448