Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Current and future climatic regions favourable for a globally introduced wild carnivore, the raccoon Procyon lotor


Invasive species are considered as one of the major threats to biodiversity and represent a major challenge in the conservation of natural ecosystems, in preventing damage to agricultural production, and human health risks. Environmental Niche Modelling has emerged as a powerful tool to predict the patterns of range expansion of non-native species and to direct effective strategies for managing biological invasions. The raccoon, Procyon lotor, is a wild mesocarnivore presenting a high adaptability and showing successful introduced populations worldwide. Here, we modelled the current and future climatically favourable areas for the raccoon using two protocols, based on data sets filtrated in geographic and environmental spaces. Projections from these models show extensive current favourable geographical areas covering extensive regions of temperate biomes. Moreover, predictions for 2050 reveals extensive new favourable areas north of the current favourable regions. However, the results of the two modeling approaches differ in the extent of predicted favourable spaces. Protocols using geographically filtered data present more conservative forecasts, while protocol using environmental filtration presents forecasts across greater areas. Given the biological characteristics and the ecological requirements of a generalist carnivore such as the raccoon, the latter forecasts appears more relevant and should be privileged in the development of conservation plans for ecosystems.


The comprehension of the relationships between organisms and their environment, and an understanding of the dynamics of species distributions is crucial to be able to predict species responses to the current environmental crisis. The recent development of computational and statistical tools allowed the emergence of new techniques to model ecological niches and potential species distributions (e.g. refs1,2,3,4,5,6,7). By associating geo-referenced occurrence data and environmental variables, Environmental Niche Modelling (hereafter called ENM) (also known as Species Distribution Modelling), makes it possible to evaluate the potential ecological niche of a species, and thus to identify the geographical areas that are favourable to its presence8,9,10,11. Associated with future climate change scenarios, ENMs also make it possible to predict the evolution of these favourable areas within a temporal framework12,13,14,15,16,17,18,19. In this way, ENMs open up a wide range of research, especially with regard to biodiversity conservation issues, notably through the study of rare or threatened species20,21,22,23,24,25,26,27,28,29,30, or conversely, of invasive alien species (e.g. refs31,32,33).

Invasive alien species are now considered as one of the major threats to biodiversity and represent a prime challenge in the conservation of natural ecosystems. The present study focuses on a globally introduced mammal, the northern raccoon (Procyon lotor). The raccoon is a carnivore native to the North American continent. Yet, this species has been moved to different areas through the pet trade and for commercial purposes (exploitation of its fur). These activities have led to its introduction and spread in many regions worldwide. The species has been present in several Caribbean islands since the 17th century34, in Japan since the 1960s35, in Azerbaijan and Iran since 199136, but also in several European countries where populations are currently expanding37. Since its introduction in the 1930s in Hessen, Germany, the species has quickly dispersed to bordering countries, and other independent introduction events have also been reported. The raccoon was observed in France as early as 1934, in the Netherlands in 1960, in Austria in 1974, in Switzerland in 1976, and in Luxemburg in 197938. Raccoons have also been observed in Denmark, Belgium, Czech Republic, Poland38, and very recently in Spain39 and Italy40,41.

The raccoon appears well adapted to urban environments, which increases the connectivity between occupied natural habitats, and may favour its successful establishment in introduced regions. The expansion of populations may also be favoured by an extremely versatile diet37,42, which also weakens the impact of potential competitors. The species may represent a threat, particularly to vulnerable ecosystems such as insular environments. Known to impact marine turtle43 and bird populations44 on several islands in its native distribution range, the raccoon has often been suspected to impact bird45, turtle and iguana populations46,47 of several Caribbean islands where the species has been introduced. However, its influence as a competitor or predator in native ecosystems, notably in continental environments, remains poorly documented37,48, and the impact of its introductions remains to be rigorously assessed. The raccoon is also considered as an agricultural pest, responsible of extensive damage to crops, orchards, and livestock feed49,50. Finally, the species is a recognized vector of diseases such as rabies and nematode-mediated pathologies51,52,53,54,55,56, raising concern for wildlife managers and agricultural producers.

ENMs have quickly emerged as important tools in identifying biotic and abiotic environmental factors that may influence the spread and distribution of non-native species16,57,58,59,60,61,62. The use of ENMs in invasive alien species studies has greatly contributed to their development63,64, and also highlighted various constraints and difficulties inherent to this technique65,66,67. While some issues lie in the quality of the data (taxonomic identification errors, sampling bias, geo-referencing precision), major difficulties arise from methodological limitations, such as the selection of the spatial framework of study, model-based variations and uncertainty, or the choice of methods for model evaluation68,69,70,71,72,73. Moreover, most ENM techniques require absence data for model calibration. Yet, these data are often difficult to obtain and are seldomly available. Hence, virtual absences, called pseudo-absences, have been proposed as an alternative. These pseudo-absences are randomly generated over the entire available environmental space74. However, this method presents a risk of generating false absences, largely impacting model performance and the identification of favourable areas75. Thus, different alternative methods have been explored, generally relying on the application of buffer distances from known presence localities76,77,78,79. Recently, methods relying on the identification of the environmental space occupied by the species, in order to generate pseudo-absences in the unoccupied environmental space, have demonstrated significant improvement of model performance80.

Another methodological difficulty associated with ENMs, particularly in invasive species studies, lies in the choice of data used for the modelling. Early approaches suggested the use of native occurrences in model calibration, and projection of these models into invaded regions81. This approach relies on the principle of niche conservatism, which assumes the retention of inherited niche-related ecological traits over time and space82. However, this assumption has been highly debated over the past decade, as niche shifts have been reported in invasive plants (e.g. refs83,84,85), insects86,87, reptiles88,89, amphibians90, and birds91. However, some authors suggested that these niche shifts might result more from a partial representativeness of the fundamental niche in the native range (for example due to dispersal capacities, competition, predation, or the absence of all favourable climatic conditions in the native area), rather than from true evolutionary niche shifts92,93,94. Alternatively, some authors suggested that niche conservatism could be restricted to short-to-moderate time spans and a lower taxonomic ranks95, while niche shifts might occur at evolutionary timescales, with variation among higher taxonomic ranks96,97. Another idea is that particular dimensions (e.g. temperature, precipitations, annual and seasonal variations) could be more conserved than others, and that niche conservatism may not encompass the entire niche98,99,100,101. Although this debate has not yet reached a consensus, considering niche conservatism or not is particularly important regarding the selection of data used in modelling. In fact, if niche conservatism is not considered, modelling favourable spaces of an introduced species requires model calibration using both native and non-native range occurrences58,102,103. Moreover, considering that niche shifts are common in invasive species has profound consequences on the interpretation of potential distributions predicted by ENMs94,104,105. Consequently, evaluation of niche conservatism represents a crucial step in invasive species niche modelling103,106.

In this study, our first objective is to compare the bioclimatic space occupied by the raccoon in its native range versus the three non-native areas where it has substantially spread (the Caribbean region, Europe, and Japan). Our second objective is to model the current and future climatically favourable areas on the basis of environmental niche models. To this end, we applied two modelling protocols, a first one based on a geographical filtration of species presences and a random selection of pseudo-absences, and a second one based on an environmental filtration of species presence and selection of pseudo-absences outside the environmental space occupied by the species. We applied an ensemble modelling procedure based on nine statistical models, two future scenarios based on the extreme ends of the Representative greenhouse gas Concentration Pathways (RCP) scenarios, and a consensus of three Global Circulation Models (GCM).


Environmental space comparisons

Our results indicated very little overlap between the native niche and either of the Caribbean, the European, and the Japanese niches (Caribbean: D = 0.01; European: D = 0.08; Japanese: D = 0.16; Table 1; Fig. 1). Moreover, niche similarity was rejected as D indices fell outside the 95% confidence interval of simulated values (Table 1). However, niche equivalence was rejected only for the Caribbean niche. In addition, we found that in all invaded areas, niches presented low expansion, but high unfilling compared to native niches, indicating that the species has not colonized all the possible environmental conditions shared with the native niche or that those conditions were absent from the non-native range (Table 1; Fig. 1).

Table 1 Tests of niche overlap, niche equivalency and niche similarity between the native niche and the niches of the three main non-native regions of Procyon lotor.
Figure 1

Analyses of environmental space shifts between the native range and the three main regions of introduction of Procyon lotor. (a) Environmental occupancy of P. lotor in native and non-native ranges along the two first axes of the PCA. Grey to black gradients in native range graph represent the density of occurrence. Solid lines represent 100% of the available environmental space. Dashed lines represent 50% of the environmental space. (b) Contribution of the bioclimatic variables to the two first axes of the PCA, and percentage of inertia explained by the two axes. (c) Occupancy of the niche along each variable gradient. A solid red arrow represents the shift of the non-native niche along the variable gradient. A dashed red arrow represents the shift of the available conditions in the non-native range. Solid lines represent 100% of available environmental space.

Bioclimatic niche model

Modelled niches differed between the EF and GF approach, with sharper responses modelled for all variables except bio1 under the EF approach (Fig. 2). Most importantly, the EF approach identified a clear response to bio2 and bio12, whereas the GF approach identified very shallow responses. Conversely, EF niches showed a broader tolerance to bio1 (a plateau between 8 and 20°C) than GF niches (a single optimum at 8 °C).

Figure 2

Response curves of the favourability value predicted by the models used in the models with Geographic Filtration (blue) and Environmental Filtration (red) approaches. Dashed blue and red lines represent favourability threshold identified for predictions of respectively GF and EF approaches. Solid blue and red lines represent the favourable range along the variable gradient.

As expected, both EF and GF approaches predicted similar favourable areas for the species over most of the temperate and sub-tropical regions (Figs 3, 4 and S1). However, the EF approach also predicted substantially larger favourable areas than GF, representing respectively 44.2% and 11.9% of all terrestrial areas (Figs 3, 4 and S1). Interestingly, the EF approach predicted that most tropical areas in Africa, and South-East Asia were favourable for the raccoon, whereas GF results were limited to temperate and dry areas of the world.

Figure 3

Projection of global bioclimatic favourability for Procyon lotor, predicted through Geographical Filtration and Environmental Filtrations approaches.

Figure 4

Projection of global bioclimatic favourable areas for Procyon lotor, predicted through Geographical Filtration and Environmental Filtrations approaches. Pink circles represent occurrences used in the modelling.

Both approaches present important differences in favourable areas at 2050. Favourable spaces are predicted to respectively increase by 18.5% (with RCP2.6; Supplementary Fig. S2) or 17.6% (with RCP8.5; Fig. 5) of its actual size with the GF approach, with broad new areas increasing from the northern border of the native range and in north-east Europe. Alternatively, the EF approach predicted a slight contraction of favourable areas by 2050, varying between 2.1% (RCP2.6; Supplementary Fig. S2) to 5.5% (RCP8.5; Fig. 5) of actual favourable areas. Lost spaces are distributed along tropical and sub-tropical regions, resulting in a shift of favourable spaces toward the northern regions of North America and Eurasia. Finally, both approaches predict that currently occupied areas are preserved in 2050.

Figure 5

Predicted favourable range change for Procyon lotor by 2050 according to scenario RCP8.5. Unfavourable: areas that are currently unfavourable remain unfavourable in the future; Lost: areas currently favourable that will lose their favourable nature in the future; Maintained: areas that are currently favourable and will still be favourable in the future; New: areas that are currently not favourable but would become favourable in the future.


This study presents the most complete analysis of the bioclimatic envelope favourable to the presence of the globally introduced carnivore Procyon lotor, and the first forecast of areas favourable to the species at a global scale using multiple modelling techniques and methods. Our results highlight the variability in environmental conditions between the native range and the regions where the raccoon was introduced. Consequently, niche models were computed with occurrences sampled from both native and non-native areas. These models created using two distinct methods demonstrate the high variability in results arising from presence selection and pseudo-absence generation techniques. Accordingly, projections for 2050 reveal major differences in favourable space from one model to another. However, both approaches highlight the ecological plasticity of the raccoon, showing large spaces favourable to the species at a global scale. Furthermore, projections for 2050 predict a high stability of current favourable areas, particularly in regions where the raccoon is already present.

Niche conservatism tests performed for the three main regions colonized by the raccoon showed contrasting results regarding niche equivalence and niche similarity. Despite the fact that equivalency tests suggest that the European and the Japanese niches are relatively conserved, similarity tests reveals significant differences between the three regions of introduction and the native niche. Yet, a common factor appears to explain the differences between the native niche and the three colonized areas. As shown in several other invasive species107,108, these differences are explained more by unfilling than by expansion. Unfilling may be the result of different processes including ongoing colonization, slow dispersal, or the impossibility to reach new areas98,105,109,110. Unfilling values for the Caribbean and Japan niches may then result from the peculiar dispersion and distribution constraints involved by insular environments. On the other hand, unfilling values for the European region can be explained, at least in part, by the ongoing expansion of the populations.

The differences in environmental conditions between the different areas where the species is present, particularly in the Caribbean region, show the necessity of taking into account occurrence data from both native and non-native areas in order to model the bioclimatic niche of the raccoon as accurately as possible. Moreover, considering that the niche of a species is conserved from a region to another implies, in absolute terms, that the niches are at equilibrium, i.e. that the species is present in all the available niches in the introduced area, and absent in all regions not in accordance with its ecological requirements. However, regarding the raccoon, as the European populations are still expanding, the possibility of niche expansion in later stages of colonization cannot be excluded111. In that respect, the niche of the raccoon cannot be considered conserved in the different regions of introduction.

In this study, two modelling approaches were used. These two approaches differ by the selection of occurrence data and by the method of generation of pseudo-absences. The first approach, GF (Geographical Filtration), relies on using occurrence data that are spatially thinned and pseudo-absences that are randomly generated within all the available environmental space. The second approach, EF (Environmental Filtration), is based on an environmental filtration of occurrence data and the use of pseudo-absences randomly generated outside of the environmental space occupied by the species.

Consequently, model responses differ substantially. Overall, response curves obtained through the EF approach display a better normalization and a reduced variance. This results in projections diverging particularly with regard to the extent of regions identified as favourable. As expected, the GF approach generates smaller favourable areas than the EF approach. This difference may arise from the fact that model responses for bio1 present a substantially wider range of favourable temperatures with the EF than with the GF approach. In addition, the geographical filtration, particularly subject to sampling biases, maintains an artificial heterogeneity in the representation of the favourable environmental conditions that may also explain the differences between the predictions of the two approaches. Finally, the random generation of pseudo-absences within the entire available geographical space inevitably leads to the generation of pseudo-absences in sites presenting known favourable environmental conditions. For example, pseudo-absences can be selected within the current range of the species. This results in a confounding influence on the identification of favourable conditions by the models, which might be more important in the case of species presenting a tolerance to a broad range of bioclimatic conditions such as the raccoon. The EF approach, by selecting pseudo-absences within conditions where the species has never been reported, eliminates this bias and allows for a better characterization of the species niche.

Results obtained through the EF approach also appear to be better matched to the ecology of the species. The raccoon is present in a wide range of habitats, elevations and climatic conditions. Raccoon populations thrives in regions as diverse as the Caribbean mangroves, the European temperate forests or the Laurentian mixed forests of North America. Thus, the restricted temperature optimum as well as the limited favourable areas identified through the GF approach appear inconsistent regarding the ecological requirements of an ubiquitous mammal such as the raccoon. Moreover, the binary projections, resulting from the GF approach, and highlighting the highly favourable space, present a very incomplete coverage of the regions where the presence of the raccoon is proven, particularly in its native distribution range. Through the example of the raccoon, it appears that an approach based on a geographical filtration of the presence/pseudo-absence data can lead to an underestimation of the area favourable for the studied species; which may have important consequences in a context of biodiversity management, and particularly in regions of introduction of alien species.

Our results reveal the raccoon’s tolerance to a very wide range of bioclimatic conditions, resulting in vast areas favourable to the species at a global scale. By highlighting the regions that are most likely to be colonized, these projections represent important tools in orienting and optimizing the monitoring efforts of the species. Since the first introductions in Europe in the early 20th century, raccoon populations have thrived and are currently present in all Western European countries. Some populations are also rapidly expanding from the center of Europe to the east, where our results identify highly favourable areas, extending in to Russia and the Middle East. Similarly, our results indicate highly favourable areas covering a large number of islands, such as different islands of the Caribbean and the Japanese archipelago, but also Madagascar, New Zealand, or Tasmania. Island environments are particularly vulnerable to the colonization of invasive alien species, notably carnivores, which are often poorly represented in native ecosystems112. The impact of introduced raccoons may thus be more damaging on islands where native fauna has been little exposed to predation, and where raccoons themselves are exposed to little predation and competition.

Invasive species management is a priority for the conservation of ecosystems and biodiversity. ENMs represent particularly effective tools to grasp the distribution of these species at different spatial and temporal scales as they can take into account the ongoing and future environmental change. Our analyses of favourable geographic space to the raccoon at the horizon 2050 predict, with both approaches, a global stability of current favourable areas (Fig. 5), with the exception of the Caribbean islands where a potential extirpation of the species may take place by 2050. Moreover, new favourable areas appear to extend broadly in northern regions in continuity with maintained favourable spaces. These results were expected given the fact that climate change scenarios predict a global warming of arctic regions, and that temperature was identified as the most influential bioclimatic factor for the raccoon (three out of the four environmental variables used in modelling correspond to temperature measures; Supplementary Fig. S3). However, while lost space is limited and fragmented in GF approach projections, it appears more pronounced with the EF approach. Thus, projections with the GF approach predict a clear expansion of favourable areas, whereas projections with the EF approach show a slight contraction and a northward shift of favourable spaces. Lost spaces are principally located at tropical latitudes, and newly favourable areas mainly cover regions to the north of current favourable areas. These observations are similar for both representative greenhouse gas concentration pathways scenarios (RCP2.6 and RCP8.5), but more pronounced in the case of RCP8.5. In accordance with the EF approach predictions, a northern shift has also been predicted for the species in its native range, as is the case for other North-American carnivores113 and northern European species114,115. In northern Europe, the development of spaces that are favourable to the raccoon may represent an additional threat to boreal ecosystems, where temperature changes are predicted to be the highest116. Arctic species are exposed to fewer competitors, predators, parasites and diseases117. Potentially affected by climate change, these species can be even more vulnerable to the introduction of new competitors, predators and potential vectors of disease such as the raccoon. This risk appears all the more likely as carnivores might be better able to keep pace with climatic changes than other mammals118.

In conclusion, the raccoon presents a tolerance to a very wide range of bioclimatic conditions resulting in extensive regions currently favourable to the species. Moreover, predictions for 2050 reveals wide newly favourable areas north of the current favourable regions. However, differences in the extent of predictions between the two modelling approaches used in this study reveals the importance of the selection of presence and pseudo-absence data. This study demonstrates the importance of data selection and processing in ENM modelling given their crucial role in our understanding of species distribution dynamics and in the development of conservation plans for ecosystems.


Species occurrence data

Occurrences of P. lotor were compiled from the online databases VertNet and GBIF, in addition to the databases of the Guadeloupe National Park, Martinique Regional Park, Saint-Martin Natural Reserve, French national wildlife organizations, recent literature, and personal observations. Only recent observations (from 1950 to the present day) with complete and precise locations were selected, giving a total of 20425 occurrences (Supplementary Table S1). All occurrence records were aggregated into 0.08° cells corresponding to the resolution of environmental variables, resulting in a total number of 5922 records (Supplementary Fig. S4).

Environmental data

Current and future favourable environmental envelopes for P. lotor were calculated using 19 bioclimatic variables (derived from temperature and precipitation measures; Table 2), averaged for the period 1950–2000 from the WorldClim 1.4 database119. A 5 arc-minute spatial resolution (approximately 9 kilometers at the equator) was selected for all bioclimatic variables.

Table 2 Codes and descriptions of the bioclimatic variables and General Circulation Models selected for use in our models.

Future favourable envelopes for P. lotor were modeled using climate projections from global climate models (GCM) based on the Coupled Model Inter comparison Project Phase 5 (CMIP5) averaged for the period 2041–2060. Three GCMs were used (CNRM-CM5; GISS-E2-R; MIROC-ESM-CHEM; Table 2), for two Representative greenhouse gas concentration pathways scenarios (RCP): the most optimistic RCP2.6 (with a radiative forcing of +2.6 W/m² for the period 2000–2100), and the most pessimistic RCP8.5 (with a radiative forcing of +8.5 W/m²).

Collinearity and the influence of the 19 bioclimatic variables were tested, using a protocol adapted from Leroy et al.25 and Bellard et al.61 (Supplementary Method S1). This allowed for the identification of four non-collinear variables that significantly influenced the distribution of P. lotor (Table 2; Supplementary Figs S3 and S5).

Analysis of niche conservatism

Following the methodology proposed by Warren et al.3, and further developed by Broennimann et al.103, we calculated niche overlap, equivalence, and similarity between the native range (4956 occurrences) and the three major areas colonized by P. lotor: the Caribbean (39 occurrences), Europe (868 occurrences), and Japan (56 occurrences). Niche overlap was calculated using a PCA approach and calibrated with the four selected environmental variables for each area of interest. PCA scores of the species occurrences on the first two axes were projected onto a grid of cells delineated by the minimum and maximum PCA scores of the environmental variables. Next, a kernel density function was applied to estimate a density of occurrence for each cell of the grid (see Di Cola et al.120 for methodological details). Thereafter, niche overlap was estimated using Schoener’s D121. This index varies between 0 and 1; 0 meaning no overlap, while 1 meaning identical niches. This index was then used to assess niche equivalence and similarity. Occurrences in each compared range were pooled and randomly split into two datasets with equal size as the original dataset. Niche overlap was then estimated with the D index. This procedure was repeated 100 times to create a null distribution. The observed D was then compared to these simulated values, and the null hypothesis of equivalence between niches was rejected when observed values fell outside of the 95% confidence interval of the simulated distribution. Alternatively, niche similarity test investigates whether the niches in the native or invaded range predict one another better than expected by chance. Occurrences in one area were randomly reallocated within their respective available environmental space and simulated niches were compared to the niche of the other area with both indices. This procedure was repeated 100 times to create a null distribution. The observed D was then compared with the simulated values, and the null hypothesis of similarity of the tested niche to the other was rejected when observed values fell outside of the 95% confidence interval of the simulated distribution.

Along with niche overlap test, niche expansion and unfilling were also calculated. The expansion index corresponds to environmental conditions in the invaded area that are absent in the native area. Conversely, unfilling refers to environmental conditions in the native area that are absent in the invaded area. Both indices vary between 0 and 1.

All niche conservatism tests were performed using the package ecospat v2.1.1122 implemented in the R software123

Data preparation and pseudo-absence selection

The preparation of presence-only data for ENMs included two important steps. The first one consisted of filtering presence data to reduce autocorrelation and sampling bias (e.g., refs124,125). The second one consisted of selecting pseudo-absence data to calibrate models. Here, we applied two different methods to filter presence data and select pseudoabsences: the first method focused on the geographical space (hereafter Geographical Filtration, GF) whereas the second one focused on the environmental space (hereafter Environmental Filtration, EF).

Geographical filtration (GF) is commonly applied in the ENM literature. We used the R package SPTHIN v.2.1–2125 to remove duplicated occurrences in a radius of 100 km, and therefore reduced auto-correlation biases. Consequently, 682 occurrences were retained. Next, we randomly sampled an identical number of pseudo-absences across the available geographical space, with three repetitions. This approach does not filter out repeated occurrences in similar climatic conditions, and samples pseudo-absences in areas that can fall within areas favourable to the species. Therefore, we expect this method to produce a modelled distribution that fits more closely the initial distribution of species occurrences, a “lower-bound” estimate of the species distribution as opposed to the Environmental Filtration method described below.

Environmental filtration (EF) focuses on the distribution of data in the space of environmental variables. Varela et al.124 showed that removing duplicated records in the environmental space consistently improved the quality of model predictions, contrary to geographical filtering. Therefore, we created a gridded environmental space on the basis of the four selected environmental variables in which we projected all the conditions existing in the geographical space. Next, we projected within this environmental space all the presence points of the species and removed all duplicate points per cell, which resulted in a total of 1036 occurrences. To select pseudo-absences within the environmental space, we applied a procedure to avoid selecting pseudo-absences within environmental conditions that are favourable for the species. To do so, we calculated the restricted n-dimensional convex hull of presences, defined as the smallest convex hull encompassing all occurrence points. This restricted convex hull is considered as a proxy of the favourable environmental conditions outside which we randomly generated 1036 pseudo-absences (i.e., in equal number to the filtered occurrences) with three repetitions. Such a procedure is supported by the statistical theory of model-based designs, also known as “D-designs” which are assumed to minimize prediction variance (see Hengl et al.76). Overall, the environmental filtering of presences will result in a decreased autocorrelation124 and the convex hull will minimize the risk to sample pseudo-absences inside favourable conditions. Therefore, we expect this procedure to produce a larger estimate of the species niche than the GF procedure, hence a larger potential distribution. In addition, we consider this approach to produce a better description potential distribution of the species compared to the GF approach as it corrects for biases linked to heterogeneity in sampling intensity and reduces the risks of generating pseudoabsences falling within favourable areas.

Ensemble modelling process

Model calibration. Nine different modelling techniques were calibrated and evaluated126,127,128,129: Artificial Neural Networks – ANN130 (Ripley, 1996); Classification Tree Analysis - CTA131, Flexible Discriminant Analyses – FDA132; Generalised Additive Models – GAM133; Generalised Boosted Models – GBM134; Generalised Linear Models – GLM135; Multivariate Adaptative Regression Splines – MARS136; MAXimum ENThropy – MAXENT1; Random Forests – RF137. For each set of presence/pseudo-absence, model calibration was realized with 70% of all data. The remaining 30% were used for model evaluations. Models were calibrated and evaluated three times per set of presence/pseudo-absence. Model calibrations were realized with the R package BIOMOD 2 v3.3–74.

Because our procedure is a presence/pseudo-absence procedure, we did not calculate discrimination capacity metrics (e.g. the area under the receiver operating characteristic curve or the true skill statistics) because (1) these metrics are designed to be calculated on real absences, and (2) they are dependent on prevalence, which can lead to spurious evaluations of ENMs138,139. Rather, we evaluated our models with the Boyce index, specifically developed for such data109,140,141. The Boyce index assess how much model predictions match the observed distribution of species occurrences. Values of the Boyce index vary between −1 and +1. Positive values indicate a model with predictions that are consistent with the distribution of occurrences in the evaluation dataset whereas negative values indicate a model with predictions that are not consistent with the distribution of occurrences. Boyce index values close to zero mean that the model is not different from a random model. Models with a mean Boyce index higher than 0.7 were selected (Supplementary Fig. S6). Consequently, all models were kept for EF modelling, but MAXENT was removed for GF modelling. These analyses were performed using the R package ecospat v2.1.1122.

Forecasts of current and future favourable areas were obtained by ensemble forecasting method. Current and future ensemble forecasts represent consensual projections of the six modelling techniques, obtained through averaged distributions of favourability scores4,142,143. To discriminate current favourable and non-favourable areas, and visualize shifts by 2050, probability distributions were transformed into binary projections. These binary projections were obtained using a probability threshold that maximized the Sørensen value139. Sørensen indices were calculated using 70% of occurrences as training data, and the remaining 30% as testing data.

Data Availability

Occurrence data are available in Supplementary Table S1 of Supplementary Informations. All climate GIS layers are available as raster grids from the Worldclim database:


  1. 1.

    Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maxim entropy modeling of species geographic distributions. Int. J. Glob. Environ. Issues 231–259, (2006).

    Article  Google Scholar 

  2. 2.

    Rangel, T. F. L. V. B., Diniz-Filho, J. A. F. & Bini, L. M. Towards an integrated computational tool for spatial analysis in macroecology and biogeography. Glob. Ecol. Biogeogr. 15, 321–327 (2006).

    Article  Google Scholar 

  3. 3.

    Warren, D. L., Glor, R. E. & Turelli, M. Environmental niche equivalency versus conservatism: Quantitative approaches to niche evolution. Evolution (N. Y). 62, 2868–2883 (2008).

    Google Scholar 

  4. 4.

    Thuiller, W., Lafourcade, B., Engler, R. & Araújo, M. B. BIOMOD - A platform for ensemble forecasting of species distributions. Ecography (Cop.). 32, 369–373 (2009).

    Article  Google Scholar 

  5. 5.

    Monahan, W. B. A mechanistic niche model for measuring species’ distributional responses to seasonal temperature gradients. PLoS One 4 (2009).

    ADS  PubMed  PubMed Central  Article  CAS  Google Scholar 

  6. 6.

    Muscarella, R. et al. ENMeval: An R package for conducting spatially independent evaluations and estimating optimal model complexity for Maxent ecological niche models. Methods Ecol. Evol. 5, 1198–1205 (2014).

    Article  Google Scholar 

  7. 7.

    Hattab, T. et al. A unified framework to model the potential and realized distributions of invasive species within the invaded range. Divers. Distrib. 23, 806–819 (2017).

    Article  Google Scholar 

  8. 8.

    Guisan, A. & Thuiller, W. Predicting species distribution: Offering more than simple habitat models. Ecol. Lett. 8, 993–1009 (2005).

    Article  Google Scholar 

  9. 9.

    Elith, J. et al. Novel methods improve prediction of species’ distributions from occurrence data. Ecography (Cop.). 29, 129–151 (2006).

    Article  Google Scholar 

  10. 10.

    Elith, J. & Leathwick, J. R. Species Distribution Models: Ecological Explanation and Prediction Across Space and Time. Annu. Rev. Ecol. Evol. Syst. 40, 677–697 (2009).

    Article  Google Scholar 

  11. 11.

    Guisan, A. et al. Predicting species distributions for conservation decisions. Ecol. Lett. 16, 1424–1435 (2013).

    PubMed  PubMed Central  Article  Google Scholar 

  12. 12.

    Skov, F. & Svenning, J.-C. Potential impact of climatic change on the distribution of forest herbs in Europe. Ecography (Cop.). 27, 366–380 (2004).

    Article  Google Scholar 

  13. 13.

    Araújo, M. B., Pearson, R. G., Thuiller, W. & Erhard, M. Validation of species-climate impact models under climate change. Glob. Chang. Biol. 11, (1504–1513 (2005).

    Google Scholar 

  14. 14.

    Beaumont, L. J., Hughes, L. & Poulsen, M. Predicting species distributions: Use of climatic parameters in BIOCLIM and its impact on predictions of species’ current and future distributions. Ecol. Modell. 186, 250–269 (2005).

    Article  Google Scholar 

  15. 15.

    Thuiller, W., Lavorel, S., Araujo, M. B., Sykes, M. T. & Prentice, I. C. Climate change threats to plant diversity in Europe. Proc. Natl. Acad. Sci. 102, 8245–8250 (2005).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  16. 16.

    Hulme, P. E. Climate change and biological invasions: evidence, expectations, and response options. Biol. Rev. 92, 1297–1313 (2017).

    PubMed  Article  PubMed Central  Google Scholar 

  17. 17.

    Mohammadi, S., Ebrahimi, E., Shahriari Moghadam, M. & Bosso, L. Modelling current and future potential distributions of two desert jerboas under climate change in Iran. Ecol. Inform, (2019).

  18. 18.

    Shirk, A. J. et al. Southwestern white pine (Pinus strobiformis) species distribution models project a large range shift and contraction due to regional climatic changes. For. Ecol. Manage. 411, 176–186 (2018).

    Article  Google Scholar 

  19. 19.

    Franco, J. N. et al. The ‘golden kelp’ Laminaria ochroleuca under global change: Integrating multiple eco-physiological responses with species distribution models. J. Ecol. 106, 47–58 (2018).

    Article  Google Scholar 

  20. 20.

    Uǧurtaş, I. H., Yildirimhan, H. S. & Sevinç, M. Distribution of the Gekkonidae species in southeast Anatolia, Turkey, and new localities. Turkish J. Zool. 31, 137–141 (2007).

    Google Scholar 

  21. 21.

    Carroll, C. Role of climatic niche models in focal-species-based conservation planning: Assessing potential effects of climate change on Northern Spotted Owl in the Pacific Northwest, USA. Biol. Conserv. 143, 1432–1437 (2010).

    Article  Google Scholar 

  22. 22.

    Engler, R., Guisan, A. & Rechsteiner, L. An improved approach for predicting the distribution of rare and endangered species from occurrence and pseudo-absence data. Appl. Ecol. 41, 263–274 (2004).

    Article  Google Scholar 

  23. 23.

    Norris, K. Managing threatened species: the ecological toolbox, evolutionary theory and declining-population paradigm. J. Appl. Ecol. 41, 413–426 (2004).

    Article  Google Scholar 

  24. 24.

    Kujala, H., Moilanen, A., Araújo, M. B. & Cabeza, M. Conservation Planning with Uncertain Climate Change Projections. PLoS One 8 (2013).

  25. 25.

    Leroy, B. et al. Forecasted climate and land use changes, and protected areas: The contrasting case of spiders. Divers. Distrib. 20, 686–697 (2014).

    Article  Google Scholar 

  26. 26.

    Monnet, A. C., Hardouin, L. A., Robert, A., Hingrat, Y. & Jiguet, F. Evidence of a link between demographic rates and species habitat suitability from post release movements in a reinforced bird population. Oikos 124, 1089–1097 (2015).

    Article  Google Scholar 

  27. 27.

    Guo, J. L. et al. Conservation strategy for aquatic plants: endangered Ottelia acuminata (Hydrocharitaceae) as a case study. Biodivers. Conserv, (2019).

    Article  Google Scholar 

  28. 28.

    Bosso, L. et al. Nature protection areas of Europe are insufficient to preserve the threatened beetle Rosalia alpina (Coleoptera: Cerambycidae): evidence from species distribution models and conservation gap analysis. Ecol. Entomol. 43, 192–203 (2018).

    Article  Google Scholar 

  29. 29.

    Taylor, A. T., Papeş, M. & Long, J. M. Incorporating fragmentation and non-native species into distribution models to inform fluvial fish conservation. Conserv. Biol. 32, 171–182 (2018).

    PubMed  Article  Google Scholar 

  30. 30.

    Kabir, M. et al. Habitat suitability and movement corridors of grey wolf (Canis lupus) in Northern Pakistan. PLoS One 12, e0187027 (2017).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  31. 31.

    Gallien, L., Douzet, R., Pratte, S., Zimmermann, N. E. & Thuiller, W. Invasive species distribution models - how violating the equilibrium assumption can create new insights. Glob. Ecol. Biogeogr. 21, 1126–1136 (2012).

    Article  Google Scholar 

  32. 32.

    Ihlow, F. et al. Impacts of climate change on the global invasion potential of the African clawed frog Xenopus laevis. PLoS One 11 (2016).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  33. 33.

    Liang, W. et al. Predicting the potential invasion of kudzu bug, Megacopta cribraria (Heteroptera: Plataspidae), in North and South America and determining its climatic preference. Biol. Invasions 1–15, (2018).

    Article  Google Scholar 

  34. 34.

    Helgen, K. M., Maldonado, J. E., Wilson, D. E. & Buckner, S. D. Molecular Confirmation of the Origin and Invasive Status of West Indian Raccoons. J. Mammal. 89, 282–291 (2008).

    Article  Google Scholar 

  35. 35.

    Okabe, F. & Agetsuma, N. Habitat Use by Introduced Raccoons and Native Raccoon Dogs in a Deciduous Forest of Japan. J. Mammal. 88, 1090–1097 (2007).

    Article  Google Scholar 

  36. 36.

    Farashi, A., Kaboli, M. & Karami, M. Predicting range expansion of invasive raccoons in northern Iran using ENFA model at two different scales. Ecol. Inform. 15, 96–102 (2013).

    Article  Google Scholar 

  37. 37.

    Salgado, I. Is the raccoon (Procyon lotor) out of control in Europe? Biodivers. Conserv. 27, 2243–2256 (2018).

    Article  Google Scholar 

  38. 38.

    Kauhala, K. Introduced carnivores in Europe with special reference to central and northern Europe. Wildlife Biol. 2, 197–204 (1996).

    Article  Google Scholar 

  39. 39.

    Garcia, J. T. et al. Recent invasion and status of the raccoon (Procyon lotor) in Spain. Biol. Invasions 14, 1305–1310 (2012).

    Article  Google Scholar 

  40. 40.

    Mori, E. et al. The masked invader strikes again: The conquest of Italy by the Northern raccoon. Hystrix 26, 1–5 (2015).

    Google Scholar 

  41. 41.

    Boscherini, A., Mazza, G., Menchetti, M. & Laurenzi, A. Time is running out! Rapid range expansion of the invasive northern raccoon in central Italy. Mammalia 17–20 (2019).

  42. 42.

    Kays, R. et al. Camera Traps as Sensor Networks for Monitoring Animal Communities. The 34th IEEE Conference on Local Computer Networks. 1, 811–818 (2009).

    Google Scholar 

  43. 43.

    Garmestani, A. S. & Percival, H. F. Raccoon Removal Reduces Sea Turtle Nest Depredation in the Ten Thousand Islands of Florida. Eagle Hill Inst. 4, 469–472 (2005).

    Google Scholar 

  44. 44.

    Ellis, J. C. et al. Impact of Raccoons on Breeding Success in Large Colonies of Great Black-Backed Gulls and Herring Gulls. Waterbirds Int. J. Waterbird Biol. 30, 375–383 (2007).

    Google Scholar 

  45. 45.

    William H. The urgent need for conservation taxonomy in the bahamas: New bird species as an example. Bahamas Nat. J. Sci. (2006).

  46. 46.

    Hayes, W. K., Carter, R. L., Cyril, S. & Thornton, B. Iguanas: biology and conservation. (2004).

  47. 47.

    Laurent, O. Conservation de l’iguane des Petites Antilles (Iguana delicatissima) en Martinique: Suivi des populations sur l’îlet Chancel et réintroduction sur l’îlet Ramiers. Université Paul Sabatier, ONCFS, 1–80 (2006).

  48. 48.

    Lutz, W. The introduced raccoon Procyon lotor population in Germany. Wildlife Biol. 2, 228 (1996).

    Article  Google Scholar 

  49. 49.

    Ikeda, T., Asano, M., Matoba, Y. & Abe, G. Present Status of Invasive Alien Raccoon and its Impact in Japan. Glob. Environ. Res. 8, 125–131 (2004).

    Google Scholar 

  50. 50.

    Beasley, J. C. & Rhodes, O. E. Jr. Relationship between raccoon abundance and crop damage. Human-wildlife Conflicts 2, 248–259 (2008).

    Google Scholar 

  51. 51.

    Arjo, W., Fisher, C., Armstrong, J., Johnson, D. & Boyd, F. Monitoring raccoon rabies in Alabama: the potential effects of habitat and demographics. Wildl. Damage Manag. Confrences Proc. 96, 14–22 (2005).

    Google Scholar 

  52. 52.

    Bartoszewicz, M., Okarma, H., Zalewski, A. & Szczęsna, J. Ecology of the Raccoon (Procyon lotor) from Western Poland. Ann. Zool. Fennici 45, 291–298 (2008).

    Article  Google Scholar 

  53. 53.

    Puskas, R. B. et al. Raccoon (Procyon lotor) Movements and Dispersal Associated with Ridges and Valleys of Pennsylvania: Implications for Rabies Management. Vector-Borne Zoonotic Dis, (2010).

    PubMed  Article  PubMed Central  Google Scholar 

  54. 54.

    Beasley, J. C., Beatty, W. S., Atwood, T. C., Johnson, S. R. & Rhodes, O. E. A comparison of methods for estimating raccoon abundance: Implications for disease Vaccination programs. J. Wildl. Manage. 76, 1290–1297 (2012).

    Article  Google Scholar 

  55. 55.

    Vos, A., Ortmann, S., Kretzschmar, A. S., Köhnemann, B. & Michler, F. The raccoon (Procyon lotor) as potential rabies reservoir species in Germany: A risk assessment. Berl. Munch. Tierarztl. Wochenschr. 125, 228–235 (2012).

    PubMed  PubMed Central  Google Scholar 

  56. 56.

    Beltrán-Beck, B., García, F. J. & Gortázar, C. Raccoons in Europe: Disease hazards due to the establishment of an invasive species. Eur. J. Wildl. Res. 58, 5–15 (2012).

    Article  Google Scholar 

  57. 57.

    Gallien, L., Münkemüller, T., Albert, C. H., Boulangeat, I. & Thuiller, W. Predicting potential distributions of invasive species: Where to go from here? Divers. Distrib. 16, 331–342 (2010).

    Article  Google Scholar 

  58. 58.

    Jiménez-Valverde, A. et al. Use of niche models in invasive species risk assessments. Biol. Invasions 13, 2785–2797 (2011).

    Article  Google Scholar 

  59. 59.

    Diez, J. M. et al. Will extreme climatic events facilitate biological invasions? Front. Ecol. Environ. 10, 249–257 (2012).

    Article  Google Scholar 

  60. 60.

    Bellard, C. et al. Will climate change promote future invasions? Glob. Chang. Biol. 19, 3740–3748 (2013).

    ADS  PubMed  PubMed Central  Article  Google Scholar 

  61. 61.

    Bellard, C., Leroy, B., Thuiller, W., Rysman, J. F. & Courchamp, F. Major drivers of invasion risks throughout the world. Ecosphere 7, 1–14 (2016).

    Article  Google Scholar 

  62. 62.

    Early, R. et al. Global threats from invasive alien species in the twenty-first century and national response capacities. Nat. Commun. 7 (2016).

  63. 63.

    Araújo, M. B. & Guisan, A. Five (or so) challenges for species distribution modelling. J. Biogeogr. 33, 1677–1688 (2006).

    Article  Google Scholar 

  64. 64.

    Sax, D. F. et al. Ecological and evolutionary insights from species invasions. Trends Ecol. Evol. 22, 465–471 (2007).

    PubMed  Article  PubMed Central  Google Scholar 

  65. 65.

    Pearson, R. G. & Dawson, T. P. Bioclimate envelope models: what they detect and what they hide - response to Hampe. Glob. Ecol. Biogeogr. 13, 471–473 (2004).

  66. 66.

    Elith, J. & Graham, C. H. Do they? How do they? WHY do they differ? on finding reasons for differing performances of species distribution models. Ecography (Cop.). 32, 66–77 (2009).

    Article  Google Scholar 

  67. 67.

    Guillera-Arroita, G. et al. Is my species distribution model fit for purpose? Matching data and models to applications. Glob. Ecol. Biogeogr. 24, 276–292 (2015).

    Article  Google Scholar 

  68. 68.

    Allouche, O., Tsoar, A. & Kadmon, R. Assessing the accuracy of species distribution models: Prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. 43, 1223–1232 (2006).

    Article  Google Scholar 

  69. 69.

    Pearson, R. G. et al. Model-based uncertainty in species range prediction. J. Biogeogr. 33, 1704–1711 (2006).

    Article  Google Scholar 

  70. 70.

    Dormann, C. F. et al. Methods to account for spatial autocorrelation in the analysis of species distributional data: A review. Ecography (Cop.). 30, 609–628 (2007).

    Article  Google Scholar 

  71. 71.

    Fernandez, M. et al. Locality Uncertainty and the Differential Performance of Four Common Niche-Based Modeling. Techniques. Biodivers. Informatics 6, 36–52 (2009).

    Google Scholar 

  72. 72.

    Anderson, R. P. Harnessing the world’s biodiversity data: Promise and peril in ecological niche modeling of species distributions. Ann. N. Y. Acad. Sci. 1260, 66–80 (2012).

    ADS  PubMed  Article  PubMed Central  Google Scholar 

  73. 73.

    Yackulic, C. B. et al. Presence-only modelling using MAXENT: When can we trust the inferences? Methods Ecol. Evol. 4, 236–243 (2013).

    Google Scholar 

  74. 74.

    Barbet-Massin, M., Jiguet, F., Albert, C. H. & Thuiller, W. Selecting pseudo-absences for species distribution models: how, where and how many? Methods Ecol. Evol. 3, 327–338 (2012).

    Article  Google Scholar 

  75. 75.

    Anderson, R. P. & Raza, A. The effect of the extent of the study region on GIS models of species geographic distributions and estimates of niche evolution: Preliminary tests with montane rodents (genus Nephelomys) in Venezuela. J. Biogeogr. 37, 1378–1393 (2010).

    Article  Google Scholar 

  76. 76.

    Hengl, T., Sierdsema, H., Radović, A. & Dilo, A. Spatial prediction of species’ distributions from occurrence-only records: combining point pattern analysis, ENFA and regression-kriging. Ecol. Modell. 220, 3499–3511 (2009).

    Article  Google Scholar 

  77. 77.

    Mateo, R. G., Croat, T. B., Felicısimo, A. M. & Munoz, J. Profile or group discriminative techniques? Generating reliable species distribution models using pseudo-absences and target-group absences from natural history collections. Divers. Distrib. 16, 84–94 (2010).

    Article  Google Scholar 

  78. 78.

    Norris, D., Rocha-Mendes, F., Frosini de Barros Ferraz, S., Villani, J. P. & Galetti, M. How to not inflate population estimates? Spatial density distribution of white-lipped peccaries in a continuous Atlantic forest. Anim. Conserv. 14, 492–501 (2011).

    Article  Google Scholar 

  79. 79.

    Bedia, J., Herrera, S. & Gutiérrez, J. M. Dangers of using global bioclimatic datasets for ecological niche modeling. Limitations for future climate projections. Glob. Planet. Change 107, 1–12 (2013).

    ADS  Article  Google Scholar 

  80. 80.

    Iturbide, M. et al. A framework for species distribution modelling with improved pseudo-absence generation. Ecol. Modell. 312, 166–174 (2015).

    Article  Google Scholar 

  81. 81.

    Peterson, A. T. Predicting the geography of species invasions via ecological niche modeling. Q. Rev. Biol. 78, 419–433 (2003).

    PubMed  Article  PubMed Central  Google Scholar 

  82. 82.

    Peterson, A. T., Soberón, J. & Sánchez-Cordero, V. Conservatism of ecological niches in evolutionary time. Science (80-.). 285, 1265–1267 (1999).

    CAS  Article  Google Scholar 

  83. 83.

    Broennimann, O. et al. Evidence of climatic niche shift during biological invasion. Ecol. Lett. 10, 701–709 (2007).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  84. 84.

    Gallagher, R. V., Beaumont, L. J., Hughes, L. & Leishman, M. R. Evidence for climatic niche and biome shifts between native and novel ranges in plant species introduced to Australia. J. Ecol. 98, 790–799 (2010).

    Article  Google Scholar 

  85. 85.

    Atwater, D. Z., Ervine, C. & Barney, J. N. Climatic niche shifts are common in introduced plants. Nat. Ecol. Evol, (2017).

    Article  Google Scholar 

  86. 86.

    Fitzpatrick, M. C., Weltzin, J. F., Sanders, N. J. & Dunn, R. R. The biogeography of prediction error: Why does the introduced range of the fire ant over-predict its native range? Glob. Ecol. Biogeogr. 16, 24–33 (2007).

    Article  Google Scholar 

  87. 87.

    Medley, K. A. Niche shifts during the global invasion of the Asian tiger mosquito, Aedes albopictus Skuse (Culicidae), revealed by reciprocal distribution models. Glob. Ecol. Biogeogr. 19, 122–133 (2010).

    Article  Google Scholar 

  88. 88.

    Rödder, D. & Lötters, S. Explanative power of variables used in species distribution modelling: An issue of general model transferability or niche shift in the invasive Greenhouse frog (Eleutherodactylus planirostris). Naturwissenschaften 97, 781–796 (2010).

    ADS  PubMed  Article  CAS  PubMed Central  Google Scholar 

  89. 89.

    Rödder, D., Schmidtlein, S., Veith, M. & Lötters, S. Alien invasive slider turtle in unpredicted habitat: A matter of niche shift or of predictors studied? PLoS One 4 (2009).

  90. 90.

    Tingleya, R., Vallinotob, M., Sequeirac, F. & Kearneyd, M. R. Realized niche shift during a global biological invasion. Proc. Natl. Acad. Sci. 111, 10233–10238 (2014).

    ADS  Article  CAS  Google Scholar 

  91. 91.

    Stiels, D., Gaißer, B., Schidelko, K., Engler, J. O. & Rödder, D. Niche shift in four non-native estrildid finches and implications for species distribution models. Ibis (Lond. 1859). 157, 75–90 (2014).

    Article  Google Scholar 

  92. 92.

    Soberón, J. & Peterson, A. T. Ecological niche shifts and environmental space anisotropy: a cautionary note Desplazamientos en el nicho y la anisotropía del espacio ambiental: una nota precautoria. Rev. Mex. Biodivers. 82, 1348–1355 (2011).

    Google Scholar 

  93. 93.

    Petitpierre, B., Broennimann, O., Kueffer, C., Daehler, C. & Guisan, A. Selecting predictors to maximize the transferability of species distribution models: lessons from cross-continental plant invasions. Glob. Ecol. Biogeogr. 26, 275–287 (2017).

    Article  Google Scholar 

  94. 94.

    Qiao, H., Escobar, L. E. & Peterson, T. A. Accessible areas in ecological niche comparisons of invasive species: Recognized but still overlooked. Sci. Rep. 7, 1–9 (2017).

    Article  CAS  Google Scholar 

  95. 95.

    Peterson, A. T. Ecological niche conservatism: A time-structured review of evidence. J. Biogeogr. 38, 817–827 (2011).

    Article  Google Scholar 

  96. 96.

    Duran, A., Meyer, A. L. S. & Pie, M. R. Climatic niche evolution in new world monkeys (Platyrrhini). PLoS One 8, 8–13 (2013).

    Google Scholar 

  97. 97.

    Waterson, A. M. et al. Modelling the climatic niche of turtles: A deep-time perspective. Proc. R. Soc. B Biol. Sci. 283 (2016).

  98. 98.

    Sánchez-Fernández, D., Lobo, J. M., Millán, A. & Ribera, I. Habitat type mediates equilibrium with climatic conditions in the distribution of Iberian diving beetles. Glob. Ecol. Biogeogr. 21, 988–997 (2012).

    Article  Google Scholar 

  99. 99.

    Sunday, J. M., Bates, A. E. & Dulvy, N. K. Thermal tolerance and the global redistribution of animals. Nat. Clim. Chang. 2, 686–690 (2012).

    ADS  Article  Google Scholar 

  100. 100.

    Araújo, M. B. et al. Heat freezes niche evolution. Ecol. Lett. 16, 1206–1219 (2013).

    PubMed  Article  Google Scholar 

  101. 101.

    Liu, X. et al. Realized climatic niches are conserved along maximum temperatures among herpetofaunal invaders. J. Biogeogr. 44, 111–121 (2017).

    Article  Google Scholar 

  102. 102.

    Broennimann, O. & Guisan, A. Predicting current and future biological invasions: Both native and invaded ranges matter. Biol. Lett. 4, 585–589 (2008).

    PubMed  PubMed Central  Article  Google Scholar 

  103. 103.

    Broennimann, O. et al. Measuring ecological niche overlap from occurrence and spatial environmental data. Glob. Ecol. Biogeogr. 21, 481–497 (2012).

    Article  Google Scholar 

  104. 104.

    Early, R. & Sax, D. F. Climatic niche shifts between species’ native and naturalized ranges raise concern for ecological forecasts during invasions and climate change. Glob. Ecol. Biogeogr. 23, 1356–1365 (2014).

    Article  Google Scholar 

  105. 105.

    Guisan, A., Petitpierre, B., Broennimann, O., Daehler, C. & Kueffer, C. Unifying niche shift studies: Insights from biological invasions. Trends Ecol. Evol. 29, 260–269 (2014).

    PubMed  Article  Google Scholar 

  106. 106.

    Fernández, M. & Hamilton, H. Ecological niche transferability using invasive species as a case study. PLoS One 10, 1–17 (2015).

    Google Scholar 

  107. 107.

    Strubbe, D., Broennimann, O., Chiron, F. & Matthysen, E. Niche conservatism in non-native birds in Europe: Niche unfilling rather than niche expansion. Glob. Ecol. Biogeogr. 22, 962–970 (2013).

    Article  Google Scholar 

  108. 108.

    Polidori, C., Nucifora, M. & Sánchez-Fernández, D. Environmental niche unfilling but limited options for range expansion by active dispersion in an alien cavity-nesting wasp. BMC Ecol. 18, 36 (2018).

    PubMed  PubMed Central  Article  Google Scholar 

  109. 109.

    Petitpierre, B. et al. Climatic Niche Shifts Are Rare Among Terrestrial Plant Invaders. Science (80-.). 338, 193–193 (2012).

    ADS  Google Scholar 

  110. 110.

    Strubbe, D., Beauchard, O. & Matthysen, E. Niche conservatism among non-native vertebrates in Europe and North America. Ecography (Cop.). 38, 321–329 (2015).

    Article  Google Scholar 

  111. 111.

    Hill, M. P., Gallardo, B. & Terblanche, J. S. A global assessment of climatic niche shifts and human influence in insect invasions. Glob. Ecol. Biogeogr. 26, 679–689 (2017).

    Article  Google Scholar 

  112. 112.

    Courchamp, F., Chapuis, J. & Pascal, M. Reviews: Mammal invaders on islands: impact, control and control impact. Biol. Rev. 78, 347–383 (2003).

    PubMed  Article  PubMed Central  Google Scholar 

  113. 113.

    Pandey, R. & Papes, M. Changes in future potential distributions of apex predator and mesopredator mammals in North America. Reg. Environ. Chang. 1223–1233 (2017).

  114. 114.

    Hof, A. R., Jansson, R. & Nilsson, C. Future Climate Change Will Favour Non-Specialist Mammals in the (Sub)Arctics. PLoS One 7, 1–11 (2012).

    Article  CAS  Google Scholar 

  115. 115.

    Elmhagen, B., Kindberg, J., Hellström, P. & Angerbjörn, A. A boreal invasion in response to climate change? Range shifts and community effects in the borderland between forest and tundra. Ambio 44, 39–50 (2015).

    PubMed Central  Article  Google Scholar 

  116. 116.

    Rounsevell, M., Fischer, M., Rando, A. T.-M. & Mader, A. The IPBES regional assessment report on biodiversity and ecosystem services for Europe and Central Asia. Secr. Intergov. Sci. Platf. Biodivers. Ecosyst. Serv. Bonn, Ger. 892 (2018).

  117. 117.

    Callaghan, T. V. et al. Biodiversity, Distributions and Adaptations of Arctic Species in the Context of Environmental Change. AMBIO A J. Hum. Environ. 33, 404–417 (2004).

    Article  Google Scholar 

  118. 118.

    Schloss, C. A., Nunez, T. A. & Lawler, J. J. Dispersal will limit ability of mammals to track climate change in the Western Hemisphere. Proc. Natl. Acad. Sci. 109, 8606–8611 (2012).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  119. 119.

    Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G. & Jarvis, A. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978 (2005).

    Article  Google Scholar 

  120. 120.

    Di Cola, V. et al. ecospat: an R package to support spatial analyses and modeling of species niches and distributions. Ecography (Cop.). 40, 774–787 (2017).

    Article  Google Scholar 

  121. 121.

    Schoener, T. W. The Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49, 704–726 (1968).

    Article  Google Scholar 

  122. 122.

    Broennimann, O. et al. Package ‘ ecospat’ (2016).

  123. 123.

    R Core Team R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, (2013).

  124. 124.

    Varela, S., Anderson, R. P., García-Valdés, R. & Fernández-González, F. Environmental filters reduce the effects of sampling bias and improve predictions of ecological niche models. Ecography (Cop.). 37, 1084–1091 (2014).

    Google Scholar 

  125. 125.

    Aiello-Lammens, M. E., Boria, R. A., Radosavljevic, A., Vilela, B. & Anderson, R. P. spThin: An R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography (Cop.). 38, 541–545 (2015).

    Article  Google Scholar 

  126. 126.

    Scherrer, D., Christe, P. & Guisan, A. Modelling bat distributions and diversity in a mountain landscape using focal predictors in ensemble of small models. Divers. Distrib. 770–782, (2019).

    Article  Google Scholar 

  127. 127.

    Fernandes, R. F. et al. Species distribution models support the need of international cooperation towards successful management of plant invasions. J. Nat. Conserv. 49, 85–94 (2019).

    Article  Google Scholar 

  128. 128.

    Rathore, P., Roy, A. & Karnatak, H. Modelling the vulnerability of Taxus wallichiana to climate change scenarios in South East Asia. Ecol. Indic. 102, 199–207 (2019).

    Article  Google Scholar 

  129. 129.

    Smeraldo, S. et al. Ignoring seasonal changes in the ecological niche of non-migratory species may lead to biases in potential distribution models: lessons from bats. Biodivers. Conserv. 27, 2425–2441 (2018).

    Article  Google Scholar 

  130. 130.

    Ripley, B. D. Pattern recognition and neural networks. Cambridge University Press, (1996).

  131. 131.

    Thuiller, W. et al. Large-scale environmental correlates of forest tree distributions in Catalonia (NE Spain). Glob. Ecol. Biogeogr. 12, 313–325 (2003).

    Article  Google Scholar 

  132. 132.

    Hastie, T., Tibshirani, R. & Buja, A. Flexible Discriminant. Analysis. J. Am. Stat. Assoc. 89, 1255–1270 (1994).

    MATH  Article  Google Scholar 

  133. 133.

    Hastie, T. & Tibshirani, R. Generalized additive models. Statistical Models 249–308 (1991).

  134. 134.

    Ridgeway, G. The State of Boosting. Comput. Sci. Stat. 31, 172–181 (1999).

    Google Scholar 

  135. 135.

    McCullagh, P. & Nelder, J. A. Generalized Linear Models, (1989).

    MATH  Book  Google Scholar 

  136. 136.

    Friedman, J. H. Greedy function machine: A gradient boosting machine. Statistics (Ber). 29, 1189–1232 (2001).

    MATH  Google Scholar 

  137. 137.

    Breiman, L. Random Forest. Mach. Learn. 45, 5–32 (2001).

    MATH  Article  Google Scholar 

  138. 138.

    Lobo, J. M., Jiménez-valverde, A. & Real, R. AUC: A misleading measure of the performance of predictive distribution models. Glob. Ecol. Biogeogr. 17, 145–151 (2008).

    Article  Google Scholar 

  139. 139.

    Leroy, B. et al. Without quality presence-absence data, discrimination metrics such as TSS can be misleading measures of model performance. J. Biogeogr. 1–9, (2018).

    Article  Google Scholar 

  140. 140.

    Boyce, M. S., Vernier, P. R., Nielsen, S. E. & Schmiegelow, F. K. A. Evaluating resource selection functions. Ecol. Modell. 157, 281–300 (2002).

    Article  Google Scholar 

  141. 141.

    Hirzel, A. H., Le Lay, G., Helfer, V., Randin, C. & Guisan, A. Evaluating the ability of habitat suitability models to predict species presences. Ecol. Modell. 199, 142–152 (2006).

    Article  Google Scholar 

  142. 142.

    Araújo, M. B. & New, M. Ensemble forecasting of species distributions. Trends Ecol. Evol. 22, 42–47 (2007).

    PubMed  Article  PubMed Central  Google Scholar 

  143. 143.

    Marmion, M., Parviainen, M., Luoto, M., Heikkinen, R. K. & Thuiller, W. Evaluation of consensus methods in predictive species distribution modelling. Divers. Distrib. 15, 59–69 (2009).

    Article  Google Scholar 

Download references


We sincerely thank Jessica Thevenot, Jean-Marc Pons, Hervé Magnin, Guy Van Laere, Julien Chalifour, Baptiste Angin, Arnaud Lenoble and François Catzeflis for sharing occurrence data. We thank Tarek Hattab for his valuable advices. This project received funding from the region Ile de France (ARDoC) and from the the Investissement d’Avenir Project Labex BCDiv (ANR-10-LABX-0003).

Author information




V.L., G.V. and A.H. conceived the study. V.L. acquired the data and B.L. designed the methodology. V.L. and B.L. interpreted the results. V.L. wrote the first draft of the manuscript, and all the authors contributed to its writing.

Corresponding author

Correspondence to Vivien Louppe.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Louppe, V., Leroy, B., Herrel, A. et al. Current and future climatic regions favourable for a globally introduced wild carnivore, the raccoon Procyon lotor. Sci Rep 9, 9174 (2019).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing