Mammalia 74 (2010): 275–280 � 2010 by Walter de Gruyter • Berlin • New York. DOI 10.1515/MAMM.2010.036 2010/027 Article in press - uncorrected proof Assessment of Cerdocyon thous distribution in an agricultural mosaic, southeastern Brazil Katia Maria Paschoaletto Micchi de Barros Ferraz1,*, Marinez Ferreira de Siqueira2, Paula Sanches Martin1, Carolina Franco Esteves3 and Hilton Thadeu Zarate do Couto1 1 University of São Paulo, Luiz de Queiroz College of Agriculture, Forest Science Department, P.O. Box 09, Piracicaba, São Paulo, 13418-900, Brazil, e-mail: katia.ferraz@usp.br 2 Botanical Garden of Rio de Janeiro, 915, Pacheco Leão Street, Jardim Botânico, Rio de Janeiro, 22460-030, Brazil 3 Univ Estadual Paulista, Biosciences Institute, Department of Ecology, P.O. Box 199, Rio Claro, São Paulo, 13506- 900, Brazil *Corresponding author Abstract Crab-eating fox, Cerdocyon thous, is a habitat generalist spe- cies relatively common throughout its range inhabiting most environments owing to its low habitat requirements. Because no information is available for anthropogenic habitats, this study aimed to describe the species occurrence in a highly fragmented and heterogeneous landscape in southeastern Brazil. C. thous was surveyed in 95 study sites in four main land covers (native forest, eucalyptus forest, sugar cane and pasture) from April to September 2006. Presence records (ns28) and landscape variables (land cover, heterogeneity, stream and forest fragment distance, elevation and slope) were used for modeling in Maxent. The bootstrapping meth- od was used for sampling 70% of the dataset for training and 30% for testing models. The species was equally dis- tributed in all types of land cover, although it was more frequent in the sugar cane areas and more associated with forest fragments and heterogeneous habitats. The potential distribution model predicted forest patches and its surround- ings as highly suitable for the species. It also predicted part of the sugar cane matrix as highly suitable, probably related to prey availability. Results suggested that the anthropogenic landscape studied encompasses many suitable habitats for species occurrence, resulting in the necessity to assess its potential role in vertebrate communities. Keywords: agroecosystem; habitat generalist species; Maxent; potential distribution; track survey. Introduction The crab-eating fox, Cerdocyon thous (Linnaeus 1766) is a medium-sized nocturnal carnivore widely distributed in the Neotropical region. The species is relatively common throughout its range from Colombia and south Venezuela, into Brazil, Paraguay, northern Argentina and Uruguay (Berta 1982), inhabiting most habitats including marshland, savan- na, cerrado, caatinga, scrubland, woodland, dry and semi- deciduous forest, gallery forest and Atlantic forest, among others (Courtenay and Maffei 2008, Di Bitetti et al. 2009). Cerdocyon thous is an insectivore/omnivore and is able to use environments disturbed by human activities owing to its opportunist and generalist habits (Facure and Monteiro-Filho 1996, Facure et al. 2003, Bueno and Motta-Junior 2004, Jácomo et al. 2004, Dotta and Verdade 2007, Courtenay and Maffei 2008, Rocha et al. 2008). The generalist diet of C. thous includes fruits, plants, vertebrates and invertebrates, in addition to human rejects (Facure et al. 1996, Juarez and Marinho-Filho 2002, Jácomo et al. 2004, Rocha et al. 2004, Gatti et al. 2006a,b, Pedó et al. 2006, Rocha et al. 2008). Habitat fragmentation, usually defined as a landscape- scale process involving both habitat loss and the breaking apart of habitats, has positive and negative effects on bio- diversity (Fahrig 2003). Habitat generalist species should benefit from environments that are heterogeneous (in space and/or time) (Kassen 2002, Marvier et al. 2004). As Marvier et al. (2004) indicated, habitat destruction, fragmentation and short-term disturbances all favor invasion by habitat gener- alists. Considering the little information available for Cerdocyon thous occurrence and its potential to adapt to anthropogenic habitats, this study aimed to describe the spe- cies occurrence in a highly fragmented and heterogeneous agricultural landscape in southeastern Brazil. Ecological niche modeling was used as a tool to elucidate the potential distribution of the species in this landscape. Materials and methods Study area The study area was at the Corumbataı́ river basin (1710 km2), located in the middle-east part of São Paulo state 228049–238419 S, 478269–478569 W; Figure 1). It comprises eight municipalities and has approximately 530,000 inhabi- tants. The most important river is the Corumbataı́ river, which starts flowing in the cuesta zone, reaching Piracicaba river after crossing Rio Claro city, the most important munici- pality in the basin (Garcia et al. 2006). The topography of the region is moderate undulated. The elevation on the springs of the Corumbataı́ river is nearly 1058 m and at the discharge nearly 470 m (Garcia et al. 2006). After intensive and lasting processes of landscape modifications, approximately 12% of original Atlantic Forest Brought to you by | Universidade Estadual Paulista Júlio de Mesquita Filho Authenticated | 186.217.234.218 Download Date | 9/10/13 3:09 PM 276 K.M.P.M.B. Ferraz et al.: Crab-eating fox in an agricultural landscape Article in press - uncorrected proof Figure 1 Location of Corumbataı́ river basin in São Paulo State, southeastern Brazil with main land uses/land covers. remains in highly fragmented conditions (Valente and Vet- torazzi 2003). Only approximately 14% of forest fragments are up to 100 ha. Larger forest fragments are located mainly on the elevated areas in the west portion of the river basin. The landscape mosaic is heterogeneous encompassing mixed cultivated areas, urban areas, pasture, forest remains and eucalyptus forest. Sugar cane (;26%) and artificial pasture (;44%) are the main land uses in the Corumbataı́ river basin (Valente and Vettorazzi 2003, Figure 1). This area is under frequent and high levels of human disturbance, owing mainly to agriculture and urban development. Species survey Cerdocyon thous was surveyed in 95 study sites: 23 in sugar cane (SC), 26 in pasture (PA), 27 in native forest (NF) and 19 in eucalyptus forest (EF), selected from the stratified sam- pling approach considering the four main land covers in the river basin. After stratification systematic points were cen- tered in all map polygons by GIS software (ArcView 3.3, ESRI, CA, USA), representing possible study sites. All pos- sible study sites were checked before sampling, and those on which track survey could be feasible were selected. The spe- cies occurrence by track survey was recorded on a 350-m transects in dirt trails early in the morning (06:00–10:00 h) at once, in each study site. All tracks were measured, pho- tographed and correctly identified by mammal specialists. They were also checked according to Becker and Dalponte (1991) and Borges and Tomas (2004). All records were made by the same observer from April to September 2006. Environmental variables Ten environmental variables were considered in this study, six continuous and one categorical, with four classes of land cover (Table 1). Among these, the heterogeneity (i.e., land- scape diversity) merits some consideration. Two different levels of heterogeneity were evaluated by Shannon’s land- scape diversity index (McGarigal and Marks 1994): (a) a landscape heterogeneity map quantified by an interpolated grid by the inverse of distance weight (IDW) of systematic points 250 m distant from each other, and (b) local hetero- geneity measures for each sample site quantified by the pro- portion of land cover within a 250-m and 1000-m radius. A landscape heterogeneity map was generated to be included in the modeling and a local heterogeneity measure was cal- culated as an attribute of the sample site, considering the Brought to you by | Universidade Estadual Paulista Júlio de Mesquita Filho Authenticated | 186.217.234.218 Download Date | 9/10/13 3:09 PM K.M.P.M.B. Ferraz et al.: Crab-eating fox in an agricultural landscape 277 Article in press - uncorrected proof Table 1 Environmental variables used to describe and to predict C. thous occurrence at the Corumbataı́ river basin, southeastern Brazil. Variables Description Land cover Four categorical variable of land cover (pasture, sugar cane, vegetation, other) Landscape Landscape diversity heterogeneity Local Diversity measure within 250 m radius and heterogeneity 1000 m radius Forest fragment Gradient distance in meters from forest distance fragment Distance from Gradient distance in meters from the closest streams main stream Elevation Elevation in meters Slope Terrain slope (%) surrounding diversity. The distance of 250 m between sys- tematic points for map interpolation was considered adequate to generate the landscape heterogeneity map, because shorter distances did not include the land cover variability and larger ones did not provide a good estimate of diversity values. The difference between the calculated local heterogeneity meas- ure and that estimated by the interpolation was analyzed using the t-test with no statistical difference between mean values (ts-0.03, gls1998, ps0.975). For local heterogeneity measurements the 250-m radius was chosen to be compara- ble with the one used in the interpolation whereas the 1000- m radius was chosen to provide an area close to the one estimated as the species home range, approximately 3 km2 on average (Courtenay and Maffei 2008). Ecological niche modeling Presence records and nine variables (land cover with four classes, landscape heterogeneity, forest fragment distance, distance from stream, elevation and slope) were used for modeling in Maxent (AT&T Labs-Research, Florham Park, NJ, USA) (Phillips et al. 2006, Phillips and Dudı́k 2008) with independent data setting; 70% of the dataset for training and 30% for testing models (Pearson 2007). The datasets were sampled by the bootstrapping method with ten random par- titions with replacement. All runs were set with a conver- gence threshold of 1.0 E-5 with 500 iterations, with 10,000 background points (Maxent software default). Data analysis The observed frequency of occurrence of Cerdocyon thous was compared with that expected among land covers by the x2-test. The relation between the species presence and land- scape variables (local heterogeneity, forest fragment distance, distance from streams, elevation and slope) was clarified with a principal components analysis biplot graph (Gabriel 1971). Model performance evaluation was assessed by AUC (area under curve), a probability that a randomly chosen presence site will be ranked above a randomly chosen absence site (Phillips and Dudı́k 2008). The best model was the one with the highest AUC value. Threshold-independent assessment for the minimum value associated with the test dataset was used to avoid commission error converting the best predic- tive model into a binary map with suitable and unsuitable areas for Cerdocyon thous. Results Cerdocyon thous occurred in 29.47% (ns28) of the sampling study sites. The species was equally distributed in all land cover sampled (x2s2.778, dfs3, ps0.427), although it was more frequent in sugar cane areas (35.71%). The occurrence of C. thous was not related to any specific landscape variable (Figure 2), although 42.86% of presence records were at less than 100 m from forest fragments. Presence records were slightly associated with more heterogeneous areas near forest fragments. The potential distribution model predicted approximately 44% of the anthropogenic river basin as suitable for Cer- docyon thous occurrence (36.38% sugar cane, 27.63% pas- ture, 22.12% native forest and 12.51% eucalyptus forest), excluding urban areas, perennial fields and pasture and sugar cane far from forest area (Figure 3). Highly suitable areas for C. thous occurrence encompassed vegetation (native and eucalyptus forest) and its surroundings, and some portions of sugar cane plantations. Distance from forest fragments (35.3%) and distance from streams (25.2%) were the highest contributor variables for model prediction. The environmen- tal variable with the highest gain when used in isolation was distance from forest fragments, which therefore appears to have the most useful information by itself. This variable was also the one that most decreases the gain when omitted, sug- gesting to have the most information that is not present in the other variables. Average AUC was 0.688"0.042 and the AUC value for the best model was 0.797. Threshold assess- ment for the minimum value associated with the test dataset (0.450) with 0% test omission rate was used for discrimi- nating suitable from unsuitable areas for the species. Model validation with test subset was statistically significant (ps0.0008). Discussion Cerdocyon thous occurred in forest habitat (native and euca- lyptus forests), open habitat (pasture) and cultivated areas (sugar cane) characterizing the species as a habitat generalist and emphasizing its potential to occur in altered environ- ments. As stated by Courtenay and Maffei (2008) C. thous readily adapts to deforestation, agricultural and horticultural development (e.g., sugar cane, eucalyptus, among others) and habitats in regeneration. It is probable that this species takes advantage of using these new landscape elements for displacement and foraging. The agricultural landscape mosaic could have favored the species occurrence in this region. Habitat generalist species Brought to you by | Universidade Estadual Paulista Júlio de Mesquita Filho Authenticated | 186.217.234.218 Download Date | 9/10/13 3:09 PM 278 K.M.P.M.B. Ferraz et al.: Crab-eating fox in an agricultural landscape Article in press - uncorrected proof Figure 2 Biplot graph of principal components explaining the presence of Cerdocyon thous in relation to environmental variables at the Corumbataı́ river basin, southeastern Brazil (SC, sugar cane; PA, pasture; NF, native forest; EF, eucalyptus forest). The percentage of variance explained by axis 1 is 37.2% and by axis 2 is 25.6%. might present a tendency for expansion of range distribution, colonizing new potential habitats in dynamic landscapes (Marvier et al. 2004). They could even survive in very small patches because they can also utilize resources in their sur- roundings (Andrén 1994). Therefore, because most of the diet of sympatric carnivores such as Cerdocyon thous, Lyca- lopex vetulus (hoary fox), Procyon cancrivorus (raccoon) and Chrysocyon brachyurus (maned wolves) overlap (Juarez and Marinho-Filho 2002, Bueno and Motta-Junior 2004, Jácomo et al. 2004, Gatti et al. 2006b), ecological segrega- tion, resource competition and partitioning should be inves- tigated in anthropogenic landscapes. The higher frequency of records in sugar cane and also the high suitability of this land use for the species occurrence (as revealed by the Maxent model) is probably related to prey availability. The high abundance of small mammals in sugar cane in the same region found by Gheler-Costa (2006) could probably explain the high incidence of the species in these places. Dotta and Verdade (2007) also found higher abun- dance of Cerdocyon thous in sugar cane in the region, prob- ably owing to the same reason. The suitability of sugar cane was higher in areas closer to forest fragments. The predictive model also revealed the native and euca- lyptus forests as highly suitable for the species occurrence. The association of the species to the fragment proximity clar- ified by the biplot graph and the model suggested the impor- tance of this landscape component to the species occurrence, possibly also serving for shelter during the day. Cerdocyon thous has also been recorded in eucalyptus plantations by Lyra-Jorge et al. (2008) and Dotta and Verdade (2007). Sil- viculture, as indicated by Rocha et al. (2008), should be used by C. thous for foraging. The AUC score obtained for this model was considered satisfactory. In this valuable contribution for the large-scale model comparison Elith et al. (2006) found 64% of the best models for each species with AUC values )0.75 and an additional 14% with AUC values between 0.7 and 0.75. They agree these AUC scores indicate that predictions based on presence-only data can be sufficiently accurate to be used in conservation planning and in numerous other applications in which estimates of species distribution are relevant. Many discussions had permeated the misleading use of AUC values as an indicator of model performance (e.g., Lobo et al. 2008, Peterson et al. 2008). Despite some of the limitations imposed on the performance of ROC analysis, it has been widely used to assess the accuracy of predictive distribution models. Some explanations could justify the AUC value found in this study. First, the area being considered in this study is inside of the distribution range of Cerdocyon thous. Most of Brought to you by | Universidade Estadual Paulista Júlio de Mesquita Filho Authenticated | 186.217.234.218 Download Date | 9/10/13 3:09 PM K.M.P.M.B. Ferraz et al.: Crab-eating fox in an agricultural landscape 279 Article in press - uncorrected proof Figure 3 Maxent model of environmental suitability areas for C. thous occurrence at the Corumbataı́ river basin, southeastern Brazil (AUCs0.797 and thresholds0.450). the high accurate models (i.e., high AUC score) presented in the ecological niche literature present, in general, broad geo- graphical extent, different from the approach used here. As indicated by Lobo et al. (2008) increasing the geographical extent outside the area where the species is predicted as pres- ent entails obtaining higher AUC scores. Second, C. thous is a typical habit generalist species (Courtenay and Maffei 2008) probably leading to a lower accurate model. Some papers have demonstrated that environmentally or geograph- ically restricted species appeared to be modeled with greater accuracy than more common and generalist species (McPher- son et al. 2004, Elith et al. 2006, McPherson and Jetz 2007, Tsoar et al. 2007). Finally, the environmental gradients of the landscape could also have decreased the AUC value as a result of conflicts situations in the algorithm solution. The incorporation of landscape variables (heterogeneity and gradient distance from fragments) in ecological niche modeling was innovative and was shown to be relevant to explain the occurrence of the species in the scale considered (30 m resolution). In such circumstances, the diversity of the land use should play an important role in species occurrence, owing probably to the availability of habitats and also resources. In addition, forest fragment distance could provide useful information for the model if the distribution of points was associated with land use, but not well represented by the land use value in the presence record (e.g., presence records in the fragment edge). Carnivores can vary in their responses to fragmentation being sensitive to fragmentation (e.g., disappearing as habitat patches became smaller and more isolated), as well as being enhanced by fragmentation (e.g., increasing abundance in highly fragmented sites) or being tolerant to fragmentation (e.g., little to no effect of landscape variables on their dis- tribution and abundance) (Crooks 2002). Marvier et al. (2004) emphasize that ecological generalist species tend to be invaders because they are more successful at establishing, spreading, and attaining high population densities. The esti- mative of population size, habitat use and selection should be considered in further studies for Cerdocyon thous in the region. In conclusion, this study highlights the potential of adap- tation of Cerdocyon thous to human-made landscapes because the Maxent model predicted much of the river basin as a suitable area for the species. Results suggested that the highly fragmented and heterogeneous landscape studied encompasses many suitable habitats for C. thous occurrence raising the necessity to assess the potential role of this spe- cies in vertebrate communities. In addition, its potential for range expansion should be evaluated because landscape modification, e.g., sugar cane plantation and cattle ranching might represent new suitable habitats for foraging species. Acknowledgements We thank the Fundacão de Amparo à Pesquisa do Estado de São¸ Paulo for scholarships and financial support for field activities. We also thank the Forest Science Department (ESALQ/USP) for the logistic help to the development of this research. Silvio F.B. Ferraz helped with suggestions and data analysis. References Andrén, H. 1994. Effects of habitat fragmentation on birds and mammals in landscapes with different proportions of suitable habitat: a review. Oikos 71: 355–366. Berta, A. 1982. Cerdocyon thous. Mammalian Species 186: 1–4. Becker, M. and J.C. Dalponte. 1991. Rastros de Mamı́feros Sil- vestres Brasileiros-Um Guia de Campo. EDUNB. 180 pp. Borges, P.A.L. and W.M. Tomas. 2004. Guia de rastros e outros vestı́gios de mamı́feros do Pantanal. EMBRAPA. 148 pp. Bueno, A.A. and J.C. Motta-Junior. 2004. Food habits of two syn- topic canids, the maned wolf (Chrysocyon brachyurus) and the crab-eating fox (Cerdocyon thous), in southeastern Brazil. Revis- ta Chilena Historia Nat. 77: 5–14. Courtenay, O. and L. Maffei. 2008. Cerdocyon thous. In: IUCN 2008. 2008 IUCN Red List of Threatened Species. -www.iucnredlist.org). Downloaded on 29 April 2009. Crooks, K. 2002. Relative sensitivities of mammalian carnivores to habitat fragmentation. Conserv. Biol. 16: 488–502. Di Bitetti, M.S., Y.E. Di Blanco, J.A. Pereira, A. Paviolo and I.J. Pérez. 2009. Time partitioning favors the coexistence of sym- patric crab-eating foxes (Cerdocyon thous) and pampas foxes (Lycalopex gymnocercus). J. Mammal. 90: 479–490. Dotta, G. and L.M. Verdade. 2007. Trophic categories in a mammal Brought to you by | Universidade Estadual Paulista Júlio de Mesquita Filho Authenticated | 186.217.234.218 Download Date | 9/10/13 3:09 PM 280 K.M.P.M.B. Ferraz et al.: Crab-eating fox in an agricultural landscape Article in press - uncorrected proof assemblage: diversity in an agricultural landscape. Biota Neotrop. 7: 287–292. Elith, J., C.H. Graham, R.P. Anderson, M. Dudı́k, S. Ferrier, A. Guisan, R.J. Hijmans, F. Huettmann, J.R. Leathwick, A. Leh- mann, J. Li, L.G. Lohmann, B.A. Loiselle, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J.M. Overton, A.T. Peterson, S.J. Phillips, K.S. Richardson, R. Scachetti-Pereira, R.E. Schapire, J. Soberon, S. Williams, M.S. Wisz and N.E. Zimmermann. 2006. Novel methods improve prediction of spe- cies’ distributions from occurrence data. Ecography 29: 129– 151. Facure, K.G. and E.L.A. Monteiro-Filho. 1996. Feeding habits of the crab-eating fox, Cerdocyon thous (Carnivora, Canidae), in a suburban area of southeastern Brazil. Mammalia 60: 147–149. Facure, K.G., A.A. Giaretta and E.L.A. Monteiro-Filho. 2003. Food habits of the crab-eating-fox, Cerdocyon thous, in an altitudinal forest of the Mantiqueira range, southeastern Brazil. Mammalia 67: 503–511. Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol. Systemat. 34: 487–515. Gabriel, K.R. 1971. The biplot graphic display of matrices with applications to principal component analysis. Biometrika 58: 453–467. Garcia, G.J., S.L. Antonello and M.G.M. Magalhães. 2006. The environmental atlas of the Corumbataı́ Watershed-SP, Brazil. Rev. Brasil. Cartograf. 58: 73–79. Gatti, A., R. Bianchi, C.R.X. Rosa and S.L. Mendes. 2006a. Diet of the crab-eating fox, Cerdocyon thous (Carnivora, Canidae) in Paulo Cesar Vinha State Park, Espı́rito Santo State, Brazil. Mam- malia 21: 153–155. Gatti, A., R. Bianchi, C.R.X. Rosa and S.L. Mendes. 2006b. Diet of two sympatric carnivores, Cerdocyon thous and Procyon can- crivorus, in a restinga area of Espı́rito Santo State, Brazil. J. Trop. Ecol. 22: 227–230. Gheler-Costa, C. 2006. Distribuicão e abundância de pequenos¸ mamı́feros em relacão à estrutura da paisagem: a sub-bacia do¸ Rio Passa-Cinco como modelo. Dissertation. Escola Superior de Agricultura Luiz de Queiroz, University of São Paulo. Jácomo, A.T., L. Silveira and A.F. Diniz-Filho. 2004. Niche sepa- ration between the maned wolf (Chrysocyon brachyurus), the crab-eating fox (Dusicyon thous) and the hoary fox (Dusicyon vetulus) in Central Brazil. J. Zool. Lond. 262: 99–106. Juarez, K.M. and J. Marinho-Filho. 2002. Diet, habitat use, and home ranges of sympatric canids in central Brazil. J. Mammal. 83: 925–933. Kassen, R. 2002. The experimental evolution of specialists, gener- alists, and the maintenance of diversity. J. Evol. Biol. 15: 173–190. Lobo, J.M., A. Jiménez-Valverde and R. Real. 2008. AUC: A mis- leading measure of the performance of predictive distribution models. Global Ecol. Biogeogr. 17: 145–151. Lyra-Jorge, M.C., G. Ciocheti and V.R. Pivello. 2008. Carnivore mammals in a fragmented landscape in northeast of São Paulo State, Brazil. Biodivers. Conserv. 17: 1573–1580. Marvier, M., P. Kareiva and M.G. Neubert. 2004. Habitat destruc- tion, fragmentation, and disturbance promote invasion by habitat generalists in a multispecies metapopulation. Risk Anal. 24: 869–878. McGarigal, K. and B.J. Marks. 1994. Fragstats spatial pattern anal- ysis program for quantifying landscape structure. Version 2.0. Forest Science Department, Oregon State University. McPherson, J.M. and W. Jetz. 2007. Effects of species’ ecology on the accuracy of distribution models. Ecography 30: 135–151. McPherson, J.M., W. Jetz and D.J. Rogers. 2004. The effects of species’ range sizes on the accuracy of distribution models: eco- logical phenomenon or statistical artifact? J. Appl. Ecol. 41: 811–823. Pearson, R.G. 2007. Species’ distribution modeling for conservation educators and practitioners. Synthesis. American Museum of Natural History. Available at http://ncep.amnh.org. Pedó, E., A.C. Tomazzoni, S.M. Hartz and A.U. Christoff. 2006. Diet of crab-eating fox, Cerdocyon thous (Linnaeus) (Carnivora, Canidae), in a suburban area of southern Brazil. Rev. Brasil. Zool. 23: 637–641. Peterson, A.T., M. Papes and J. Soberón. 2008. Rethinking receiver operating characteristic analysis applications in ecological niche modeling. Ecol. Model. 213: 63–72. Phillips, S.J. and M. Dudı́k. 2008. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31: 161–175. Phillips, S.J., R.P. Anderson and R.E. Schapired. 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190: 231–259. Rocha, V.J., N.R. Reis and M.L. Sekiama. 2004. Dieta e dispersão de sementes por Cerdocyon thous (Linnaeus) (Carnı́vora, Cani- dae), em um fragmento florestal no Paraná, Brasil. Rev. Brasil. Zool. 21: 871–876. Rocha, V.J., L.M. Aguiar, J.E. Silva-Pereira, R.F. Moro-Rios and F.C. Passos. 2008. Feeding habits of the crab-eating Fox, Cer- docyon thous (Carnivora: Canidae), in a mosaic area with native and exotic vegatation in Southern Brazil. Rev. Brasil. Zool. 25: 594–600. Tsoar, A., O. Allouche, O. Steinitz, D. Rotem and R. Kadmon. 2007. A comparative evaluation of presence-only methods for model- ling species distribution. Divers. Distrib. 13: 397–405. Valente, R.O.A. and C.A. Vettorazzi. 2003. Mapeamento de uso e cobertura dos solos da Bacia do Rio Corumbataı́, SP. Circular Técnica do IPEF 196, Piracicaba. Brought to you by | Universidade Estadual Paulista Júlio de Mesquita Filho Authenticated | 186.217.234.218 Download Date | 9/10/13 3:09 PM