Species composition and frequency of habitat use by medium and large-sized mammals in the Brazilian Cerrado Biome, State of Tocantins

The Cerrado biome is the second largest natural vegetation type existing in Brazil after the Amazon, covering 91% of the State of Tocantins, with 30% of the territory changed as a result of the expansion of economic development areas. The substitution of natural vegetation means habitat loss on species, and may put the survival of medium and large-sized mammal species under extinction risk. These species represent significant energy demands and high requirements of area within the habitat. This study aims to assess the species composition and the frequency of use of physiognomic types of vegetation by medium and large-sized mammals. The study was conducted from 2001 to 2011 using a set of complementary and alternative techniques in fragments of seven different physiognomic types of vegetation. Fourth-seven medium and large-sized mammals were registered. This record corresponds to 92% of species already described within the biome, and five additional species to those registered in the data base of the State of Tocantins. Fourteen of these species are at certain degree of threat of extinction. The most used physiognomic types of vegetation in terms of canopy cover were Cerradao, Typical Cerrado, Riparian Forest and Dense Cerrado, and the less used were Gallery Forest, Campo Sujo and Vereda.


Introduction
Cerrado is the second largest of Brazil's major habitat types, after the Amazonian rainforest, and accounts 24% of the country's land area (Instituto Brasileiro de Geografia e Estatística [IBGE], 2004;2012). Estimates show that this biome consists of various physiognomic types of vegetation, and covers 91% of the territory of the State of Tocantins. This is the only one of four States fully included in the implementation region of the Matopiba Plan for Farming Development, which is found to be the greater border for the expansion and intensification of the agricultural production in Brazil (Buainain, Garcia, & Vieira-Filho, 2015).
The impacts of suppression and substitution of natural vegetation mean habitat loss on species, and fragmentation, putting the survival of ecological communities under destruction and/or degradation risk (Fahrig, 2003), including medium and large-sized mammals. Theoretical and empirical evidences show that the greater energy demand of mammalian species with greater body weight means lesser density, greater life areas and, consequently, greater probability for extinction especially during the most advanced degradation stages of the habitat (Grelle, Paglia, & Silva, 2006).
The spatial variation of Cerrado, which is shown as a mosaic of vegetation types such as plantations, savannas and forestry (Ribeiro & Walter, 1998), constitutes a great diversity of habitats, a determining factor for a remarkable alternation of species, which may create conditions for studies on habitats utilization rates by animals. Considering that studies on use of Cerrado habitats by medium and large-sized mammal are scarce and punctual (Fonseca & Redford, 1984;Redford & Fonseca, 1986;Johnson, Saraiva, & Coelho, 1999;Santos-Filho & Silva, 2002;Bocchiglieri, Mendonça, & Henriques, 2010;Carmignotto & Aires, 2011), and constitute important subsidies for the preservation of species and types of vegetation, this

Area of study
The study was developed in the Cerrado biome of the State of Tocantins, Central-West Region of Brazil. The typical climate of this biome is hot and semi-humid, with pronounced seasonal variation marked by a dry winter (Aw -Köppen climate classification). Precipitation levels in the State vary widely. However, the mean annual rainfall is 1,500 mm, and the mean temperature is 24 o C.
Ten locations were sampled at North -South longitudinal axis of the State of Tocantins. Then, representative areas of fragments of Riparian Forest, Gallery Forest, Cerradao, Dense Cerrado, Typical Cerrado, Campo Sujo and Vereda, were selected as outlined in Ribeiro and Walter (1998). Figure 1 shows locations selected to the study, and Table 1 describes physiognomic types of vegetation under study.  Table 1. Locations sampled to the study and physiognomic types of vegetation selected according to Ribeiro and Walter (1998).

Sampling of mammals
The present study shows data collected from February 2001 to December 2011. Two methods were used for data collection as follows: (i) transect walk (Cullen Jr, Rudran, & Valladares-Padua, 2004), covering preexisting traces on sampling areas at 1 km hour -1 , for data collection on species by means of direct observation and observation of vestiges (indirect method); and (ii) camera traps placed on the best spots, where animal usually pass through, to capture unique images of wildlife 24 hours a day.
The sampling effort varied according to the sampling method used for data collection. The sampling effort was 1,245 hours for transect walk, and 3,646 traps a day for camera traps method (Srbek-Araujo & Chiarello, 2005).

Data processing and analysis
Species richness estimates for medium and large-sized mammal were carried out from the listing of species. The presence-absence matrix was then developed and used to estimate the species richness observed and its confidence intervals. The Jackknife estimator of species richness was used, and the EstimateS 9.1.0 (Colwell, 2013) was run based on 1,000 random samples without replacement.
The chi-squared test (χ 2 ) was carried out in PAST software (Hammer, Harper, & Ryan 2001) to compare the utilization frequency of physiognomic types of vegetation by medium and large-sized mammal species at 5% significance, assuming that p-values lesser than 0.05 show strong evidence against the null hypothesis (Triola, 1999).
The association between medium and large-sized mammal species and physiognomic types of vegetation was assessed by means of Principal Component Analysis (PCA), based on the number of records on the occurrence of species within each sampled type of vegetation. The Multivariate Statistical Package (Kovach, 2013) was used for PCA.
The Species-sampling effort curve obtained from the species richness and the Jackknife estimator of species richness showed a stabilization trends with 82.4% sampling efficiency. This achievement was expected once the study lasted long and was developed in a wide study area, using a set of complementary and alternative sampling techniques (Figure 2).
Fourth-seven medium and large-sized mammalian species were registered. Thirty-five of 47 species were registered in the Cerradao, 30 in the Typical Cerrado, 29 in the Riparian Forest, 22 in the Dense Cerrado, 10 in the Gallery Forest, 9 in the Campo Sujo and 4 in the Vereda. These data show that the species under study have using all the existing physiognomic types of vegetation, however, at different utilization rates. Figure 3 shows the Principal Component Analysis (PCA) for medium and large-sized mammal species, in which the number of records was sufficient for the analysis. The total variance in the set of variables under study was 72.5, where 43.7% for the axis 1 and 28.8% for the axis 2. Thus, the two-dimensional perceptual graph generated by the PCA was found to be appropriate to assess relationships between variables (Rencher, 2002;Hongyu, Sandanielo, & Oliveira-Jr, 2016).
The axis 1 describes positive relationships existing with species occurred in savannas such as Cerradao, Dense Cerrado, Typical Cerrado and Campo Sujo, as follows: Cabassous unicinctus (Cn), Cerdocyon thous (   The chi-squared test showed that there are no statistically significant differences regarding the use of Riparian Forest, Typical Cerrado and Dense Cerrado by medium and large-sized mammals (0.02 ≤ χ 2 ≤ 2.96, and 1.00 ≥ p-value ≥ 0.11). These were the most used habitats by mammal species under study. The Gallery Forest, Campo Sujo and Vereda were habitats with lesser frequency use when compared with other types of vegetation sampled in the study area (4.50 ≤ χ 2 ≤ 19.88, and 0.04 ≥ p-value ≥ 0.01).
The lower richness of medium and large-sized mammalian species observed in the Gallery Forest differ from those found in studies conducted in the Cerrado, which have showed that this type of vegetation contributes significantly to changes of fauna diversity in the Cerrado once is used as wildlife movement corridor, climate stability environment and high structural complexity. In addition, this habitat provides shelter, fodder and water in prolonged period of time with little or no rain (Redford & Fonseca, 1986;Johnson et al., 1999;Santos-Filho & Silva, 2002;Ribeiro & Marinho-Filho, 2005;Ribeiro & Melo, 2013). A putative explanation on this contradictory finding may be related to the number of fragments sampled in each of physiognomic type of vegetation under study. A set of ten fragments was sampled in each of Riparian Forest, Cerradao, Dense Cerrado and Typical Cerrado habitats, and four fragments were sampled in the Gallery Forest. The same may have determined the lower species richness observed for Campo Sujo and Vereda habitats, in which the species-sampling effort comprised six and three fragments, respectively. Besides, these types of vegetation are found to be habitats with lower structural complexity, therefore, with lower richness of mammalian species (Ribeiro & Marinho-Filho, 2005

Conclusion
The species composition and frequency of habitat use by medium and large-sized mammals in the Brazilian Cerrado biome varied widely. Most species under study are already described within the biome and five are registered in the data base of the State of Tocantins. Fourteen of 47 species registered are at certain degree of threat of extinction.
Medium and large-sized mammalian species showed preference to a certain physiognomic type of vegetation such as Cerradao, Typical Cerrado, Dense Cerrado and Riparian Forest. Beyond this inference, it points out that lesser frequency of species in habitats such as Gallery Forest, Campo Sujo and Vereda was observed.