Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Pet distribution modelling: Untangling the invasive potential of Trachemys dorbigni (Emydidae) in the Americas

  • Érica Fonseca ,

    Roles Conceptualization, Data curation, Formal analysis, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Departamento de Biologia, Programa de Pós-Graduação em Biodiversidade Animal, Universidade Federal de Santa Maria, Santa Maria, Rio Grande do Sul, Brazil

  • Camila Both,

    Roles Supervision, Visualization, Writing – review & editing

    Affiliation Departamento Interdisciplinar, Universidade Federal do Rio Grande do Sul, Campus Litoral Norte, Tramandaí, Brazil

  • Sonia Zanini Cechin,

    Roles Resources, Supervision, Writing – review & editing

    Affiliation Departamento de Biologia, Programa de Pós-Graduação em Biodiversidade Animal, Universidade Federal de Santa Maria, Santa Maria, Rio Grande do Sul, Brazil

  • Gisele Winck

    Roles Conceptualization, Methodology, Supervision, Validation, Writing – review & editing

    Current address: Laboratório de Biologia e Parasitologia de Mamíferos Silvestres Reservatórios, Instituto Oswaldo Cruz, Fundação Oswaldo Cruz, Rio de Janeiro, RJ, Brazil

    Affiliation Laboratoire d’Écologie Alpine (LECA), Université Grenoble Alpes, CNRS, Université Savoie Mont Blanc, Grenoble, France


Human activities have been changing the global biogeographic patterns by the introductions of invasive species. For reptiles, the invasion rate increase of non-native species is remarkably related to the pet trade, especially for freshwater turtles. Here we estimated the invasive potential of the South American turtle Trachemys dorbigni in the Americas using a combination of climatic and human activity variables. We built species distribution models based on data from the native and invasive ranges, using the ensemble model from five different algorithms (GAM, MAXENT, BRT, RF and GBM). We compared the two models’ performance and predictions, one calibrated with only climatic variables (climate-driven), and the second also included a descriptive variable of human activity (climate plus human-driven). Suitable areas for T. dorbigni covered occurrence areas of its congeners and highly diversified ecoregions, such as the eastern USA, the islands of Central America, and the south eastern and eastern Brazilian coast. Our results indicate that human activities allow T. dorbigni to establish populations outside of its original climatic niche. Including human activity variables proved fundamental to refining the results to identify more susceptible areas to invasion and to allow the efficient targeting of prevention measures. Finally, we suggested a set of actions to prevent T. dorbigni becoming a highly impacting species in the areas identified as more prone to its invasion.


Human activities continuously change the global biogeographic patterns through the establishment of introduction pathways of non-native species, increasing biotic exchange rate [1, 2]. As a consequence, biological invasions are among the main threats to biodiversity [3]. Increased numbers of invasive species are associated with the increasing number of individuals introduced and introduction events (the propagule pressure; [4, 5]). Biological invasion theory deals with the invasion process as a multistage continuum, where a high propagule pressure is essential for the successful establishment and maintenance of small populations during the early stages of invasion (sensu [6]). This high propagule pressure allows a continuous gene flow even in nonsuitable areas and, therefore, increases the probability of adaptation [7]. Specifically in the pet trade, the continuous release of individuals into nature reduces the reliance on reproductive success for long-term population maintenance, which increases the chances of establishment. The number of invasive species introduced by the legal and illegal pet trade has increased over the last decades worldwide, influenced by commercial expansion through online commerce, and especially in social networks [810]. New invasions are expected to continually emerge in the next years due to a lag phase, when species remain with few individuals in the introduction area for a long time before becoming invasive [11, 12]. Therefore, initiatives to identify the most prone areas to invasion before these nonindigenous species become invasive are critical for conserving native populations and planning for the prevention and management of invasive species.

Both biotic (e.g., competition, facilitation, vegetation cover) and abiotic (e.g., temperature, humidity) filters also influence the successful establishment of an invader species [13]. These environmental constraints are often associated with physiological limitations and demands for suitable sites for reproduction. Despite the general lack of detailed ecological data for most invasive populations, a general macroecological approach using spatial statistical models can provide important information from the available biotic and abiotic data (e.g., [1417]. Species distribution modelling (SDM) is a commonly used tool to estimate suitable areas for nonindigenous species. SDMs assist in risk assessment by identifying areas prone to invasions and provides valuable information for mitigation measures ([18, 19]. These models can be used in a more comprehensive approach by including other environmental variables, such as vegetation cover, invasive species introduction events, or transport pressure (see [15]). For invasive species modelling, the inclusion of variables that quantify human activity in SDM improves model estimation power [20, 21]. For example, exotic pet introduction depends on their release into the environment; thus, propagule pressure for these animals should be higher in large urban centres and their surroundings, and closer to residences [22, 23]. As a result, the inclusion of human activity measures in the SDM allows combining the probability of introduction or release (and accessibility) with climatic suitability of habitat [24, 25] to provide more realistic scenarios.

In this study, we used SDMs to understand the invasive risk of a South American freshwater turtle, Trachemys dorbigni (Duméril and Bibron, 1835), within the Americas. This species is native to southern Brazil, Uruguay and northern Argentina [26, 27], inhabiting different aquatic environments with abundant vegetation (lakes, slow-flowing rivers, ponds, and wetlands), and urban areas [28, 29]. Due to pet release, T. dorbigni was introduced in areas outside of its native distribution, becoming invasive at some locations in north-eastern and south-eastern Brazil. However, its population impacts are still poorly known [30, 31]. This study is the first approach on its potential for invasion, providing information to prevent future impacts on potential invaded communities and freshwater ecosystems, since one of its congener (Trachemys scripta; Thunberg in Schoepff, 1792) figures as a top-100 worst invasive species of the world. Banning T. scripta trade in Europe and in some South American countries [3234] increased pressure upon other species of freshwater turtles [34], including our targeted species. Indeed, egg and new-born removal for the pet trade is the main threat within its native area, together with water pollution and road kills [35]. Here we combine two sets of variables to model habitat suitability for Trachemys dorbigni in the Americas, one including only climatic (abiotic) variables and the second including a human activity index; this allowed us to: (1) identify which areas are more prone to T. dorbigni invasion, and (2) to analyse human activity as a major driver of increased environmental suitability for this species.


Species occurrence data

We compiled occurrence records of Trachemys dorbigni in its native range and invaded areas from the global biodiversity information facility (GBIF; [36]), the VertNet data portal (, Hórus Institute [37], and peer-reviewed published articles (S1 Appendix). Our data set included the largest number of occurrence records with a minimal pairwise distance of 10 Km to prevent sampling bias [38], using the randomisation function from ‘spThin’ R package [39]. The final dataset contained 56 occurrence records of T. dorbigni within its native range, plus 15 invasive population records. Because the species is freshwater-dependent, presence records are geographically biased towards rivers and lakes.

Climatic variables and measures of human activity

We included the 19 bioclimatic variables from CHELSA [40] and two topographic variables (elevation and slope) derived from the Hydro-1K global digital elevation model [41]. Since we were explicitly interested in how human activity influences the invasion potential of T. dorbigni, we included the global human influence index (HII; [42]) as a second data set, containing bioclimatic, topographic, and human influence variables. HII is built up from nine global data layers including three main anthropogenic pressure sources [42], human population (density), land use and infrastructure (e.g., land cover), and human accessibility (e.g., roads). All layers had a spatial resolution of 5 arc-minutes (approximately 10 Km x 10 Km at the equator).

To prevent multicollinearity, we selected the two sets of variables with the lowest correlation values based on the variance inflation factor (VIF) using the ‘fmsb’ R package [43]. VIF is an effective approach for multicollinearity assessment for more than two independent variables at a time; the higher the VIF value, the higher the collinearity between the related variables [44]. We only selected the bioclimatic variables below the critical threshold of five: isothermality (bio 3), temperature annual range (bio 7), mean temperature of the wettest quarter (bio 8), mean temperature of the warmest quarter (bio 10), and precipitation seasonality (bio 15). Therefore, one predictor dataset comprised only the climatic variables selected by VIF, and other combined the climatic variables and the human influence index.

Species distribution models

We calibrated two SDM sets for T. dorbigni: one set using only climate variables (climate-driven; CD) with both native and invasive occurences, and another set including human-related variables (climate plus human-driven; CHD), using only invasive occurrences. To assess the role of anthropogenic activities on the environmental suitability and the invasive potential of T. dorbigni, we jointly used native and invasive occurrences. We believe joining occurrences would improve its fundamental niche characterization, comparing to models with native occurrences only [18]. Likewise, we run CHD models with invaded occurrences only to incorporate anthropogenic effects on the release and/or escape of exotic species. Human activity may produce an opposite effect on the native area, negatively impacting the species instead of allow its expansion [45, 46]. We projected the mean resulting models (CD and CHD models) into wider-ranged layers (the Americas) to identify the most suitable areas for population establishment. We assumed that higher-valued suitability areas were more prone to successful invasions.

We identified the areas with higher invasive potential for the turtle using the ensemble model from five different algorithms (GAM, MAXENT, BRT, RF and GBM) through the ‘biomod2’ R package [47]. Ensemble models can provide more robust predictions for potential distributions of invasive species, where species-environment relationships are difficult to determine [48, 49]. Ensemble distribution maps highlight areas of agreement among different models predictions, reducing areas of uncertainty for individual models [50, 51].Because these algorithms require presence and absence records, and absence records are usually either unavailable or unreliable, we replaced them by pseudo-absence occurrence sets [52]. We generated a random sample set of pseudo-absences proportional to the number of presences [53] with three replicates within the study area. We set equal prevalence weight for both presence and pseudo-absence groups (0.5; [54]) to prevent model overfitting and to produce a more accurate prediction for potential invasions [53, 5557].

For model evaluation, we randomly divided the data set into 80% for calibration and 20% for validation. We repeated this procedure 10 times using 10 random data set splits, totalling 100 models for each type of model (CD and CHD). To evaluate model-fitting, we used the true skill statistic (TSS). TSS is a precision-dependent measure, insensitive to prevalence [58]. To build a consensus prediction, we only included higher-performance models in the ensemble forecasting (TSS >0.7; [59]). We compared the TSS values of the CD and CHD models using Wilcoxon’s statistics to verify if the different sets of variables affected the performance of the models. Final outputs were two continuous suitability layers comprising the mean ensemble of the best predictions for each model type (CD and CHD), both projected in the Americas. Continuous predictions were converted into binary maps by using a threshold value which maximized both sensitivity and specificity to show the differences and convergences between each model.Unlike most studies with invasive species, here we prioritised maximising specificity (the proportion of absences correctly predicted) in defining model parameters; that is, we chose predictions with higher specificities to reduce the number of false positives (true absences predicted as presence; [53]). The ensemble distribution maps represent only the areas with the greatest potential to be suitable for T.dorbigni.


As expected, in the climate plus human-driven model (CHD) suitable areas coincided with areas in which the Trachemys dorbigni is invasive (Fig 1). Using the ensemble approach, the climate-driven model (CD) identified suitable areas for T. dorbigni in north-central Argentina along the Rio Negro drainage basin, eastern USA, and the Great Basin in the USA (sensu [60]), in addition to their natural range (Fig 1A). These ecoregions are typically arid and temperate, presenting independent river systems and drainage basins [61, 62]. Other smaller areas of dry and moist forests in the south-eastern and eastern Brazilian coast, northern Andean region, Caribbean islands, Central America, and parts of Chile were identified by the CHD model (Fig 1B). A comparison between the distributions predicted by both models (CD and CHD) can be found in Fig 1C.

Fig 1. The resulting suitability areas for the successful establishment of T. dorbigni in the Americas.

(A) climate-only model (CD); (B): climate + human activity model (CHD); (C) comparison of predicted distributions in the CD and CHD models; light blue areas show where the CHD model predicts species presence, while dark blue areas areas show where the CD model predicts species presence. Areas in green indicate coincidence in model predictions. Map images hosting provided by the Center for Spatial Sciences at the university of California, Davis.

TSS values for ensemble models are 0.87 for CD model and 0.99 for CHD model, which is considered a good performance (Fig 2). However, the CHD model differed significantly from CD, with higher TSS (Wilcoxon rank-sum test on TSS (W = 16145, p = 2.895e-11). Therefore, our results from both models comprised good predictions on habitat suitability for the species, but the CHD model performed better according to model evaluation metrics.

Fig 2. TSS values for the climate-driven model and combined model (climate and human activity index).


The CHD model outperformed the CD model at predicting known invaded areas, and in model precision. This suggests that human-driven environmental modifications allow T. dorbigni to establish populations outside its original climatic niche. Although the successful establishment is influenced by propagule pressure and abiotic and biotic characteristics of the invaded area, the magnitude and extent of the effect of each factor are often modified by human activity [13]. Human activity increases the likelihood of population establishment, by increasing genetic diversity and the chances of adaptation through the constant release of pets (see [7]). At the same time, environmental modification creates favourable environments (e.g., increased local temperature, lack of vegetation cover) and facilitates the occupation of empty niches by non-native species [63, 64]. Therefore, if human activity at the site is high enough, it may favor the establishment of exotic species in areas outside the appropriate climatic range. Population establishment of invasive species in unfavourable climatic regions is not an uncommon event; for example, introduced populations of T. scripta in Italy usually survive in habitats where bioclimatic conditions are unfavourable to reproduction [65]. This may be the case for populations of T. dorbigni inhabiting areas outside of their suitable range, since most of their invasion sites are deforested areas within the Brazilian Atlantic Forest domain, where invasion risks and propagule pressure are higher, due to the influence of the 72% (ca. 145 million) of the human population Brazilian living within its domain [66]. Almost all of the sites invaded by T. dorbigni are located in the Atlantic Forest region and were all correctly predicted by the CHD model.

Suitable areas for T. dorbigni also agreed with the distribution of T. scripta in North America, as well as the distribution of other slider turtles (Trachemys sp.) in Central and South America, especially T. stejnegeri, T. decussata, T. terrapen and T. venusta (S1 Fig). These results suggest that Trachemys species occupy similar habitats and share characteristics of life history [67]; this is especially worrying for the maintenance of native congener populations, mainly because non-native reptiles are overall more likely to successfully establish where congeners are present [67]. According to Darwin’s preadaptation hypothesis, the presence of congeners indicates that resource availability and physiological tolerances are compatible with local conditions [68]. However, the relationship between congener presence and the successful invasion of turtles on a global scale has not yet been demonstrated [69]. Further analyses on a smaller scale (regional and local) may clarify the importance of congeners in the success of invasive slider turtles. In contrast, the impacts of introducing non-native slider turtles are known and include changes in behaviour and distribution of native turtle populations by biotic interactions, disease spread, and hybridisation [70, 71]. Hybridisation cases involving Trachemys species are already observed [71, 72] suggesting the lack of physiological barriers between species. In a successful invasion scenario of T. dorbigni in congener areas, the phylogenetic proximity between Trachemys species suggests that their hybridisation could occur [73, 74]. In addition, sliders are considered highly competitive and aggressive turtles in their interactions [75, 76]. Competition for food, and nesting and basking sites displaces native turtle species from the most suitable places for thermoregulation and where resources are most profitable [75, 77]. Competitive displacement, together with hybridisation, can alter population structure and endanger the survival of native populations, decreasing reproductive success, and increasing mortality rates [70, 78].

Given the stochastic nature of each invasion and the uncertainties related to the introduction of new propagules, to precisely predict which areas are most at risk becomes a challenge. Although climatic variables are important, especially for ectothermic species, they alone are not enough to describe invasive species distribution [79]. The occurrence of T. dorbigni can be also limited by other factors besides thermal tolerance [80]. As freshwater invaders, these organisms are more prone to occupy large areas by using water body networks for dispersal, together with the known pet introduction pathways. The accessibility to other lakes, ponds, and rivers may be key for the spread of this species, and difficulties for dispersal process include habitats split by dams and roads, for example. In this sense, alternative models run with the freshwater environmental database and other variables associated with the presence of water can refine future models and improve predictions by including due inclusion of likely dispersal routes for aquatic invaders [45]. Furthermore, it is likely that its occurrence is also limited by density-dependent factors (e.g., intraspecific competition, density, mortality rates) which reduces the need of individuals to disperse. Therefore, local abiotic and biotic patterns also exert pressure on the establishment and dispersion of T. dorbigni, and should be considered in future plans to contain and prevent invasion of the species.

Here we used species distribution models not only to identify areas more prone to T. dorbigni invasion, and which variables would explain better the suitability, but also to understand the role of human activities on the invasion process. Our results comply with previous studies highlighting the importance of including human activities as variables in SDMs to improve model predictions of invasive species, especially freshwater turtles [65, 79, 81, 82]. Because human activity is highly influential in the invasion process, we believe that even when calibrating models that account for the interaction between species (e.g., JSDM, Joint Species Distribution Model), the human variable should enhance the prediction power of invasion-prone areas. However, it is important to highlight that the SDMs consider that the species would be able to reach all areas equally and that the conditions will remain the same offered in the construction of the model. Thus, changes in parameters (e.g., changes in accessibility or vegetation cover) may affect these projections and, therefore, the interpretation of these results must consider such limitations.

In a nutshell, T. dorbigni invasiveness is driven by both climatic features and human activities. Our study generated important information to employ in direct conservation efforts at specific geographic areas in the Americas, since the habitat suitability for T. dorbigni includes the occurrence areas of its congeners and highly diverse ecoregions with great importance for conservation. Therefore, we believe the most imperative actions to prevent T. dorbigni from becoming a highly impacting non-native species within prone-invasive areas are: 1) limit or even ban commercial trade. Alternatively, all breeders and shops must implant a subcutaneous microchip for identification of the turtle, and the commercial establishment should register the owner in a country and/or worldwide database of non-native species. 2) Authorities and governments to set high fines for those who release individuals, and establish inspection activity on pet breeders and shops. 3) To promote citizen science programs to increase invasive species monitoring areas, including a user-friendly database to report findings and data. The threat of invasive species can only be mitigated by the close cooperation of the triad of scientists, civil society, and policymakers.

Supporting information

S1 Fig. Native distribution of the Trachemys genus in the Americas.

Black polygon: geographical distribution of T. dorbigni. Red polygons:1- T. scripta; 2- T. venusta; 3- T. decussata; 4- T. gaigeae; 5- T. grayi; 6- T. nebulosa; 7- T. ornata; 8- T. stejnegeri; 9- T. taylori; 10- T. terrapen; 11- T. yaquia; 12- T. adiutrix. Modified from Fritz (2012). Map images hosting provided by the Center for Spatial Sciences at the university of California, Davis. CD: climate-only model, CHD: climate + human activity model.


S1 Table. Geographic coordinates, states and country of occurrence records of Trachemys dorbigni.


S2 Table. Variables importance (mean values ± SD) from the climate-only (CD) and combined (climate + human activity; CHD) models for Trachemys dorbigni.

Variables: bio3 = isothermality; bio7 = temperature annual range; bio8 = mean temperature of wettest quarter; bio10 = mean temperature of warmest quarter; bio15 = precipitation seasonality; HII = human influence index.


S1 Appendix. List of published papers holding Trachemys dorbigni occurrence information.



EF is grateful to João Carlos Nabout and Micael Rosa Parreira for assisting with to use the human footprint variable.


  1. 1. Capinha C, Essl F, Seebens H, Moser D, Pereira HM. The dispersal of alien species redefines biogeography in the Anthropocene. Science. 2015; 348: 1248–1251. pmid:26068851
  2. 2. Meyerson LA, Mooney HA. Invasive alien species in an era of globalization. Front. Ecol. Environ. 2007; 5: 99–208.[199:IASIAE]2.0.CO;2.
  3. 3. Baillie JEM, Griffiths J, Turvey ST, Loh J, Collen B. Evolution Lost: Status E Trends of the World’s Vertebrates. London: Zoological Society of London; 2010.
  4. 4. Blackburn TM, Lockwood JL, Cassey P. The Influence of numbers on invasion success. Mol. Ecol. 2015; 24: 1942–1953. pmid:25641210
  5. 5. Cassey P, Delean S, Lockwood JL, Sadowski JS, Blackburn TM. Dissecting the null model for biological invasions: A meta-analysis of the propagule pressure effect. PLoS Biol. 2018; 16: e2005987. pmid:29684017
  6. 6. Blackburn TM, Pyšek P, Bacher S, Carlton JT, Duncan, RP, Jarošík V, et al. A proposed unified framework for biological invasions. Trends Ecol. Evol. 2011; 26: 333–339. pmid:21601306
  7. 7. Lockwood JL, Cassey P, Blackburn TM. The role of propagule pressure in explaining species invasions. Trends Ecol. Evol. 2005; 20: 223–228. pmid:16701373
  8. 8. Kraus F. Alien Reptiles and Amphibians: A Scientific Compendium and Analysis. New York: Springer; 2009.
  9. 9. Bush ER, Baker SE, Macdonald DW. Global trade in exotic pets 2006‐2012. Conserv. Biol. 2014; 28: 663–676. pmid:24661260
  10. 10. Auliya M, Altherr S, Ariano-Sanchez D, Baard EH, Brown C, Brown R. M. et al. Trade in live reptiles, its impact on wild populations, and the role of the European market. Biol. Conserv. 2016; 204: 103–119.
  11. 11. Essl F, Dullingerc S, Rabitsch W, Hulme PE, Hülber K, Jarošík V, et al. Socioeconomic legacy yields an invasion debt. Proc. Natl. Acad. Sci. 2011; 108: 203–207. pmid:21173227
  12. 12. Richardson DM. Fifty Years of Invasion Ecology. Oxford: Wiley-Blackwell; 2011.
  13. 13. Catford JA, Jansson R, Nilsson C. Reducing redundancy in invasion ecology by integrating hypotheses into a single theoretical framework. Divers. Distrib. 2009; 15: 22–40.
  14. 14. Pollock LJ, Tingley R, Morris WK, Golding N, O’Hara RB, Parris KM, et al. Understanding co‐occurrence by modelling species simultaneously with a Joint Species Distribution Model (JSDM). Methods Ecol. Evol. 2014; 5: 397–406.
  15. 15. Tingley R, García‐Díaz P, Arantes CRR, Cassey P. Integrating transport pressure data and species distribution models to estimate invasion risk for alien stowaways. Ecography. 2018; 41: 635–646.
  16. 16. Lake TA, Briscoe RD, Moeller DA. Species Distribution Models and Joint Species Distribution Models of Nine Invasive Species in North America; 2019 [cited 2019 Nov 5]. Database: Data Repository for the University of Minnesota [Internet]. Available from:
  17. 17. Mothes CC, Stroud JT, Clements SL, Searcy CA. Evaluating ecological niche model accuracy in predicting biotic invasions using South Florida’s exotic lizard community. J. Biogeogr. 2019; 46: 432–441.
  18. 18. Jiménez-Valverde A, Peterson AT, Soberón J, Overton JM, Aragón P, Lobo JM. Use of niche models in invasive species risk assessments. Biol. Invasions. 2011; 13: 2785–2797.
  19. 19. Barbet-Massin M, Rome Q, Villemant C, Courchamp F Can species distribution models really predict the expansion of invasive species? PLoS ONE. 2018; 13: e0193085. pmid:29509789
  20. 20. Zhu G, Li H, Zhao L. Incorporating anthropogenic variables into ecological niche modeling to predict areas of invasion of Popillia japonica. J. Pest. Sci. 2016; 1–10.
  21. 21. Mungi NA, Coops NC, Ramesh K, Rawat GS. How global climate change and regional disturbance can expand the invasion risk? Case study of Lantana camara invasion in the Himalaya. Biol. Invasions. 2018; 20: 1849–1863.
  22. 22. Molina FB. Ambientes urbanos e a fauna de repteis no Estado de São Paulo: entre desconhecidos, desaparecidos e indesejaveis. In: Glezer R, Mantovani MSM (Eds.). Parques urbanos e meio ambiente: Desafios de uso. São Paulo: Parque Cientec/USP, PEFI; 2006. pp. 203–222.
  23. 23. Anastácio PM, Ribeiro F, Capinha C, Banha F, Gama M, Filipe AF, et al. Non-native freshwater fauna in Portugal: A review. Sci. Total Environ. 2019; 650: 1923–1934. pmid:30286358
  24. 24. Gallardo B, Vila L. Human influence, key to understand the biogeography of invasive species in the Anthropocene. Cuad. de Investig. Geogr. 2018;
  25. 25. Hill MP, Binns M, Umina PA, Hoffmann AA, Macfadyen S. Climate, human influence and the distribution limits of the invasive European earwig, Forficula auricularia, in Australia. Pest. Manag. Sci. 2018; 75: 134–143. pmid:30168641
  26. 26. Bager A, Freitas TRO, Krause L. Nesting ecology of a population of Trachemys dorbigni (Emydidae) in southern Brazil. Herpetologica. 2007; 63: 56–65.
  27. 27. Alcalde L, Derocco NN, Rosset SD, Williams JD. Southern- most localities of Trachemys dorbigni and first record of Trachemys scripta elegans for Argentina (Cryptodira, Emydidae). Chelonian Conserv. Biol. 2012; 11: 128–133.
  28. 28. Bujes CS, Verrastro L. Quelônios do Delta do Jacuí, RS, Brasil: uso de hábitats e conservação. Nat. Conserv. 2008; 6: 47–60.
  29. 29. Bager A, Krause L, Freitas TRO. Fidelity to nesting sites and orientation of Trachemys dorbigni (Duméril & Bibron, 1835) (Testudines: Emydidae) female in southern Brazil. Trop. Zool. 2012; 25: 31–38.
  30. 30. Santos TG, Vasconcelos TS, Molina FB, Zaher H. First record of Trachemys dorbigni (Duméril & Bibron, 1835) (Testudines, Emydidae) in a remnant of Mesophytic Semideciduous Forest of Sao Paulo State, southeastern Brazil. Herpetol. Bull. 2009; 108: 27–30.
  31. 31. Santana DO, De-Carvalho CB, Rocha SM, Freitas EB, Faria RG. Trachemys dorbigni (Duméril & Bibron, 1835) (Testudines: Emydidae) recorded in an artificial pond in northeastern Brazil. Herpetol. Notes. 2014; 7: 211–213. pmid:28580298
  32. 32. Brasil. Ministério do Meio Ambiente. Portaria No. 93, de 7 de julho de 1998. DOU 128:1–7.
  33. 33. ANLA. Autoridad Nacional de Licencias Ambientales, Resolución 447 de 15 de mayo de 2013. Colombia: Minist. Ambient y Desarro. Sost; 2013.
  34. 34. Kopecky O, Kalous L, Patoka J. Establishment risk from pet-trade freshwater turtles in the European Union. Knowl. Managt. Aquatic. Ecosyst. 2013; 410: 02.
  35. 35. Vogt RC, Fagundes CK, Bataus YSL, Balestra RAM, Batista FRW, Uhlig VM, et al. Avaliação do Risco de Extinção de Trachemys dorbigni (Duméril & Bibron, 1835) no Brasil. Processo de avaliação do risco de extinção da fauna brasileira. ICMBio. 2015. Available from:
  36. 36. Global Invasive Species Database–GISD; 2019. [cited 2019 Oct 23]. Database: GISD [Internet]. Available from: on 08-05-2019
  37. 37. Hórus Institute. Base de dados de espécies exóticas invasoras do Brasil, Instituto Hórus de Desenvolvimento e Conservação Ambiental, Florianópolis–SC; 2016 [cited 2016 Oct 25] Database: Hórus Institute [Internet]. Available from:
  38. 38. Anderson RP, Gonzalez I Jr. Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with Maxent. Ecol. Modell. 2011; 222: 2796–2811.
  39. 39. Aiello-Lammens ME, Boria RA, Radosavljevic A, Vilela B, Anderson RP. spThin: an R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography. 2015; 38: 541–545.
  40. 40. Karger DN, Conrad O, Böhner J, Kawohl T, Kreft H, Soria-Auza RW, et al. Kessler M. Climatologies at high resolution for the earth’s land surface areas. Sci. Data. 2017; 4: 170122. pmid:28872642
  41. 41. USGS. HYDRO 1k, elevation derivative database. 2001; [cited 2019 Nov 7] Database: edcdaac [Internet]. Available from:
  42. 42. WCS—Wildlife Conservation Society and CIESIN—Center for International Earth Science Information Network at Columbia University. Last of the Wild Project, Version 2 (LWP-2): Global Human Influence Index (HII) Dataset (Geographic). Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC), 2005. Accessed 30 October 2018.
  43. 43. Nakazawa M. fmsb: Functions for Medical Statistics Book with some Demographic Data. R package version 0.6.3. 2018. Available from:
  44. 44. O’Brien RM. A Caution Regarding Rules of Thumb for Variance Inflation Factors. Qual. Quant. 2007; 41: 673–690.
  45. 45. Rodríguez-Rey M, Consuegra S, Borger L, Leaniz CG. Improving Species Distribution Modelling of freshwater invasive species for management applications. PLoS ONE. 2019; 14(6): e0217896. pmid:31206531
  46. 46. Bevan HR, Jenkins DG, Campbell TS. From pet to pest? Differences in ensemble SDM predictions for an exotic reptile using both native and nonnative presence data. Frontiers Biogeography. 2019; 11(2):
  47. 47. Thuiller W, Georges D, Engler R, Breiner F. 2019; biomod2: Ensemble Platform for Species Distribution Modeling. R package version 3.3–7.1. 2019. Available from:
  48. 48. Roura-Pascual N, Brotons L, Peterson AT, Thuiller W. Consensual predictions of potential distributional areas for invasive species: A case study of Argentine ants in the Iberian Peninsula. Biological Invasions. 2009; 11(4):1017–1031,
  49. 49. Stohlgren TJ, Ma P, Kumar S, Rocca M, Morisette JT, Jarnevich CS et al. Ensemble habitat mapping of invasive plant species. Risk Analysis. 2010; 30(2): 224–235, pmid:20136746
  50. 50. Marmion M, Parviainen M, Luoto M, Heikkinen RK, Thuiller W. Evaluation of consensus methods in predictive species distribution modelling. Diversity and Distributions. 2009; 15: 59–69.
  51. 51. Poulos HM, Chernoff B, Fuller PL, Butman D. Ensemble forecasting of potential habitat for three invasive fishes. Aquatic Invasions. 2012; 7(1): 59–72.
  52. 52. Elith J, Kearney M, Phillips S. The art of modelling range-shifting species. Methods Ecol. Evol. 2010; 1: 330–342.
  53. 53. Barbet-Massin M, Jiguet F, Albert CH, Thuiller W. Selecting pseudo- absences for species distribution models: How, where and how many? Methods Ecol. Evol. 2012; 3: 327–338.
  54. 54. McPherson J, Jetz W, Rogers DJ. The effects of species’ range sizes on the accuracy of distribution models: ecological phenomenon or statistical artefact? J. Appl. Ecol. 2004; 41: 811–823.
  55. 55. Syfert MM, Smith MJ, Coomes DA. The effects of sampling bias and model complexity on the predictive performance of MaxEnt species distribution models. PLoS One. 2013; 8: e55158. pmid:23457462
  56. 56. Anderson RP, 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. 2010; 37: 1378–1393.
  57. 57. Liang W, Papeş M, Tran L, Grant J, Washington-Allen R, Stewart S, et al. The effect of pseudo-absence selection method on transferability of species distribution models in the context of non-adaptive niche shift. Ecol. Modell. 2018; 388: 1–9.
  58. 58. Allouche O, Tsoar A, Kadmon R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. 2006; 43: 1223–1232.
  59. 59. Diniz-Filho JAF, Bini LM, Rangel TF, Loyola RD, Hof C, Nogués‐Bravo D, et al. Partitioning and mapping uncertainties in ensembles of forecasts of species turnover under climate change. Ecography. 2009; 32: 897–906.
  60. 60. Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GVN, Underwood EC, et al. Terrestrial Ecoregions of the World: A New Map of Life on Earth. BioScience. 2001; 51: 933.
  61. 61. Ringuelet R. Rasgos fundametales de la zoogeografía de la Argentina. Physis (Buenos Aires). 1961; 22: 151–188.
  62. 62. Holland B, Orians G, Adams J. Great Basin shrub steppe. 2019 Aug 10 [cited 10 August 2019]. In: World Wildlife Fundation [Internet]. Available from: Accessed 10 August 2019.
  63. 63. Didham RK, Tylianakis JM, Gemmell NJ, Rand TA, Ewers RM. Interactive effects of habitat modification and species invasion on native species decline. Trends Ecol. Evol. 2007; 22: 489–96. pmid:17673330
  64. 64. Newbold T, Hudson LN, Hill SLL, Contu S, Lysenko I, Senior RA, et al. Global effects of land use on local terrestrial biodiversity. Nature. 2015; 520: 45–50. pmid:25832402
  65. 65. Ficetola GF, Thuiller W, Padoa-Schioppa E. From introduction to the establishment of alien species: bioclimatic differences between presence and reproduction localities in the slider turtle. Divers. Distrib. 2009; 15: 108–116.
  66. 66. Fonseca É, Both C, Cechin SZ. Introduction pathways and socio‐economic variables drive the distribution of alien amphibians and reptiles in a megadiverse country. Divers. Distrib. 2019; 1–12.
  67. 67. Mahoney PJ, Beard KH, Durso AM, Tallian AG, Long AL, Kindermann RJ. Introduction effort, climate matching and species traits as predictors of global establishment success in non- native reptiles. Divers. Distrib. 2015; 21: 64–74.
  68. 68. Darwin C. On the Origin of Species. London: John Murray; 1859.
  69. 69. Ferreira RB, Beard KH, Peterson SL, Poessel SA, Callahan CM. Establishment of introduced reptiles increases with the presence and richness of native congeners. Amphib-Reptil. 2012; 33: 387–392.
  70. 70. Cadi A, Joly P. Impact of the introduction of the red‐eared slider (Trachemys scripta elegans) on survival rates of the European pond turtle (Emys orbicularis). Biodivers. Conserv. 2004; 13: 2511–2518.
  71. 71. Parham JF, Papenfuss TJ, Van Dijk PP, Wilson BS, Marte C, Schettino LR, et al. Genetic introgression and hybridization in Antillean freshwater turtles (Trachemys) revealed by coalescent analyses of mitochondrial and cloned nuclear markers. Mol. Phylogenet. Evol. 2013; 67: 176–187. pmid:23353072
  72. 72. Figuereido PI. Verificação da ocorrência de Hibridação Entre Tartaruga‐Tigre‐D’água, Trachemys dorbigni (Duméril & Bibron, 1835) e Tartaruga‐Americana, Trachemys scripta (Thunberg & Schoepff, 1792) (Testudines, Emydidae). M.Sc. Thesis, Universidade Federal do Rio Grande do Sul. 2014. Available from:
  73. 73. Fritz U, Stuckas H, Vargas-Ramırez M, Hundsdorfer AK, Maran J, Packert M. Molecular phylogeny of Central and South American slider turtles: Implications for biogeography and systematics (Testudines: Emydidae: Trachemys). J. Zool. Syst. Evol. Res. 2012; 50: 125–136.
  74. 74. Vargas-Ramírez M, del Valle C, Ceballos C P., Fritz U. Trachemys medemi n. sp. from northwestern Colombia turns the biogeography of South American slider turtles upside down. J. Zool. Syst. Evol. Res. 2017; 55: 326–339.
  75. 75. Polo-Cavia N, López P, Martín J. Aggressive interactions during feeding between native and invasive freshwater turtles. Biol. Invasions. 2010; 13: 1387–1396.
  76. 76. Polo-Cavia N, López P, Martín J. Interference competition between native Iberian turtles and the exotic Trachemys scripta. Basic Appl. Herpetol. 2014; 28: 5‒20.
  77. 77. Polo-Cavia N, López P, Martín J. Competitive interactions during basking between native and invasive freshwater turtle species. Biol. Invasions. 2009; 12: 2141–2152.
  78. 78. Pearson SH, Avery HW, Spotila JR. Juvenile invasive red-eared slider turtles negatively impact the growth of native turtles: Implications for global freshwater turtle populations. Biol. Conserv. 2015; 186: 115–121.
  79. 79. Beans CM, Kilkenny FF, Galloway LF. Climate suitability and human influences combined explain the range expansion of an invasive horticultural plant. Biol. Invasions. 2012; 14: 2067–2078.
  80. 80. Rodrigues JFM, Coelho MTP, Ribeiro BR. Predicting fundamental and realized distributions based on thermal niche: A case study of a freshwater turtle. Acta Oecol. 2018; 88:52–57.
  81. 81. Roura-Pascual N, Hui C, Ikeda T, Leday G, Richardson D, Carpintero S, et al. The relative roles of climatic suitability and anthropogenic influence in determining the pattern of spread in a global invader. Proc. Natl. Acad. Sci. 2011; 108: 220–225 pmid:21173219
  82. 82. Banha F, Gama M, Anastácio PM. The effect of reproductive occurrences and human descriptors on invasive pet distribution modelling: Trachemys scripta elegans in the Iberian Peninsula. Ecol. Modell. 2017; 360: 45–52.