Mammal richness and diversity in Serra do Facão region , Southeastern Goiás state , central Brazil

At least 251 mammal species are recorded for the Brazilan cerrado, which, therefore, is the third richest Brazilian biome. Most mammal surveys in Brazilian cerrado result from studies performed opportunistically and in short time periods. The aims of the present study were (1) provide a checklist for the mammalian fauna based on a five-year sampling in Serra do Facão region, Southeastern Goiás state; (2) compare small non-flying mammals diversity in open and forest areas and (3) compare species diversity before and after the flood caused by a hydroelectric reservoir filling. The data was gathered in 19 sampling periods, from May 2008 to September 2013. We sampled open and forest habitats and captured non-flying small mammals with Sherman and Tomahawk live traps and pitfalls; bats were sampled with mist-nets; large mammals were recorded with camera traps, and by direct observations and track surveys in field. We found 20 species of small non-flying mammals, 10 species of bats and 33 species of larger mammals. Species diversity was greater for forest than open habitats, and was also greater before than after the complete reservoir filling. About 10% of the recorded species are included in the Brazilian official list of threatened species. The total richness represents 25% of all cerrado mammal fauna, highlighting the importance of this area for regional mammal fauna conservation.


Introduction
Currently, 701 species of mammals are known in Brazil (Paglia et al. 2012).The mammal richness reported for the cerrado varies from 227 (Carmignotto et al. 2012) to 251 species (Paglia et al. 2012).About 40% cerrado mammals are bats (Chiroptera), 31% are rodents (Rodentia), and 10% are marsupials (Didelphimorphia) (Paglia et al. 2012).This diversity places the cerrado biome as the third richest for Brazilian mammals.The number of endemic cerrado mammals varies from 25 to 32 species depending on the authors (Carmignotto et al. 2012, Paglia et al. 2012).The distribution of mammal fauna in cerrado is affected by habitat heterogeneity, being approximately 16% of species exclusive to open areas, and about 29% occur exclusively in forest environments (Marinho-Filho et al. 2002).However, the endemism rate is slightly larger for open areas (56%), highlighting the relevance of both open and forested habitats for cerrado mammals conservation (Marinho-Filho et al. 2002).
The original cerrado covered approximately 2 million km 2 .However, about half of its area was already removed by human activities, and the actual deforestation persist at rates varies between 22,000 to 30,000 km 2 /year (Klink & Machado 2005).Processes of habitat loss and fragmentation, as well as hunting pressure, are among the main threats to mammalian diversity conservation (Rodrigues et al. 2002, Costa et al. 2005, Trolle et al. 2007).Changes in the environment occur in an accelerated rate, causing loss of irreplaceable habitats at local and regional scale, and even resulting in local extinctions (Whitmore & Sayer 1992, Myers et al. 2000).
In this context, inventories are essential because they provide basic information on biological diversity composition of determined areas and regions, providing arguments and justification for the conservation of remnant areas.However, the majority of fauna surveys on cerrado's mammalian fauna are performed punctually and in short time periods.
Herein we present a mammalian inventory based on five years sampling an area affected by the flooding of the reservoir of Serra do Faca ˜o Hydroelectric Plant, in southeastern Goia ´s state.Other aims of this study are to evaluate differences in richness and diversity for small non-flying mammals between ( 1) environments (open areas and forests) and (2) sampling periods (before and after the hydroelectric plant construction).We also provide reproductive ecological observations for small non-flying mammals.

Area of study
The study was carried out in Serra do Faca ˜o region, southeastern Goia ´s state (Figure 1).The Serra do Faca ˜o region is crossed by Sa ˜o Marcos River, which belongs to Rio de la Plata Basin, the second largest river basin in South America (ANA 2007).In November 2009, a dam on the Sa ˜o Marcos River was built to form the reservoir of the hydroelectric plant of Serra do Faca ˜o.This reservoir floods ca.20,000 ha, encompassing five municipalities in the state of Goia ´s: Catala ˜o, Campo Alegre de Goia ´s, Davino ´polis, Ipameri, and Cristalina -and the municipality of Paracatu, in the state of Minas Gerais.
In the region the climate is tropical (Ko ¨ppen Aw), with temperatures between 23 and 24°C and mean annual rainfall between 1600 and 1800 mm, and two well-defined seasons, one hot and rainy (October to Match) and other colder and dry (April to September) (Cardoso et al. 2015).The landscape consists of large tracts of plantation areas and pastures, in which there is a mosaic of remnant patches of native habitats, including several open cerrado physiognomies as campo limpo, campo sujo, vereda (palm swamps) , and "cerrado sensu stricto", and forest habitats, such as semi-deciduous forest, mata ciliar (riparian forest), and gallery forest.Phytophysiognomies were identified according to Oliveira-Filho & Ratter (2002).Grassland without shrubs or trees is called campo limpo, and grassland with scattered shrubs and small trees is called campo sujo.Open cerrado physiognomies over rock soil and scattered rock outcropscerrado rupestre (Lenza et al. 2011) was present in some places.The cerrado sensu stricto have trees covering more than 30% of canopy, and presents an herbaceous/grassy layer.Veredas are valley marshes where the water table reaches the surface and palms Mauritia flexuosa are common.Semideciduous forest (mesophytic seasonal forest) generally grow on spots of calcareous fertile soils; gallery forests are placed along river banks, and the tree branches cover the water course, forming a ''gallery''.Riparian forests are also placed along river banks, but the water course is not covered by trees.The sampling sites included both open areas (grassland and "cerrado sensu stricto") and forest habitats (gallery forest and semi-deciduous forest).

Data Collection
We collected data on 19 field expeditions, with quarterly intervals, from May 2008 to February 2012.In 2013, we done three complementary expeditions in March, June and September.Initially, we gathered data in 10 sampling sites placed in open areas and 10 sampling sites placed in forest habitats (Figure 1, Supplement 1).However, after the impoundment in 2009, three forest sites were flooded.The study was carried out in areas located in the municipalities of Catala ˜o and Campo Alegre de Goia ´s, state of Goia ´s (Supplement 1).
We considered small mammals all species of small rodents and marsupials with less than 2 kg (e.g.Bennett 1990, Eisenberg & Redford 1999).This group was sampled using Sherman and Tomahawk traps, and some additional data came from pitfall traps.In every sample sites (Supplement 2), we set up 21 Sherman traps (250 x 80 x 90 mm) and seven Tomahawk traps (300 x 160 x 160 mm), which remained open during three consecutive nights in each field expedition.Traps were placed on soil level and on trees and shrubs, up to 2m high as well, in order to access the greatest number of micro-habitats.Baits consisted of a mixture of peanut butter, banana, canned sardines and cornmeal.All traps were monitored every day at dawn.The sampling effort was 31,496 traps*night.
Pitfall traps consisted by stations composed by four 35-litre plastic buckets arranged in a Y-shaped disposition.All buckets were buried in the ground, placed 4m away from each other, and connected by a plastic fence (0.5m height).Five pitfall trap stations were set at each sample site (Supplement 2).Pitfall traps were kept open for ten consecutive days in each expedition.Although pitfall traps are commonly used for herpetological sampling, they often capture small terrestrial mammals (Mengak & Guynn Jr 1987, Santos Filho et al. 2008), being able to sample species that are rarely captured by more traditional methods (Monteiro-Filho & Graipel 2006, Ca ´ceres et al. 2010).During sampling periods all traps were checked every day and we performed a total sampling effort of 49,600 pitfalls*night.
Bats were recorded only in forest environments (Supplement 2).We used five to ten mist-nets (12 m length x 3 m height) in each sampling site, for a few nights (usually 1-2) per field expedition.The nets were set on forest edges, or inside them, near food sources and shelters, and/or on trails potentially used as flight route.Nets were opened at 18:00 h and closed at midnight.The total bat sampling effort, following Straube & Bianconi (2002), was of 18,144 hours*m 2 .
We identified, measured, and weighted all animals captured.We also recorded the reproductive status of all animals.We considered as reproductive all perforated, lactating, or pregnant females.Small non-flying mammals were marked with numbered earrings (National Band & Tags -Mod.1005 -1).After data collection, we released all animals at the same place they were captured.Some specimens were collected for further confirmation of identification.All collected animals were deposited in the Mammal Collection of the University of Brası ´lia (Supplement 3).All procedures (capture, handling and marking) followed the guidelines of animal care and use by the American Society of Mammalogists (Sikes & Gannon 2011).All captures and collections were made upon authorization issued by the Brazilian Institute for the Environment (IBAMA ACCTMB No. 198/2010).We used the specific literature for taxonomic identifications (Vizotto & Taddei 1973, Emmons & Feer 1997, Weksler & Bonvicino 2005, Carmignotto & Monfort 2006, Bonvicino et al. 2008, Gardner 2008, Reis et al. 2013).Confirmation of the identifications were done by comparison with material (skins and skulls) housed in the mammal collections of the Zoology Department of the Universidade de Brası ´lia and the Museu Nacional da Universidade Federal do Rio de Janeiro.
We sampled larger mammals opportunistically by tracking them while collecting or setting traps for small mammals sampling.Additionally we set a total of six to ten camera traps (Tigrinus Analog 6.0C), at some sampling site per expedition (Supplement 2).Camera traps were set in open and forest areas, and remained in operation for about ten consecutive days in each expedition.Some were set in trails, others not, but in any case they were set apart at least 1.5 km from each other.In 2013 we changed all camera traps by another digital model (Bushnell Trophy Cam HD).We also took into account direct sights, animals hit by cars and/or carcasses found, as well as indirect observations, such as tracks and feces, obtained during active diurnal and nightly surveys.The total effort performed with camera traps was 14,508 traps*hours.We did not consider photos of the same species taken in less than one hour interval for abundance counting.All species, including carcasses, tracks, and photos were identified using pertinent literature (Becker & Dalponte 1991, Oliveira & Cassaro 1999, Borges & Toma ´s 2004, Mamede & Alho 2006, Carvalho Jr & Luz 2008, De Angelo et al. 2008).

Data Analysis
We used collector's curve based on effort by expedition for checking the reliability of our mammal survey, including all species obtained by all methods.For the analysis of diversity we used the number of individuals and not total captures of small mammals.We used rarefaction curves based on abundance for compare non-flying small mammal diversity between (1) open and forest areas, (2) before and after the reservoir filling in open areas, and (3) before and after flooding in forest habitats.All comparisons were done using Coleman index on EstimateS Version 9.1.0(Colwell 2013) and tested differences using Z tests (Zar 1999).Reproductive patterns were roughly designed by the percentage of reproductive females in dry and wet seasons.Our sampling of larger mammals with camera traps did not allow the recognition of individuals and abundance estimates.However the number of records may provide an idea of abundance or activity of a given species in the area or specific habitat type.

Results
We recorded 63 species of mammals in the study area: 33 larger mammals, 20 small non-flying mammals and 10 bats (all belonging to the Phyllostomidae family) (Table 1).The collector's curve based on sampling effort by expedition for all species stabilized on the 13 th expedition (Figure 2), indicating we performed an adequate effort to sample the local richness.We recorded eight species considered as threatened with extinction (MMA 2014), such as Puma concolor and Priodontes maximus (Table 1).
The sampling effort with camera traps was equivalent to 14,508 traps*hour, which yielded 231 records of 22 species of large mammals.Photographic records of some species are presented on Supplement 4. Data from direct and indirect observations resulted in 11 additional species that were not recorded by camera traps, totaling 33 species of large mammals observed with all thechniques combined.The most speciose large mammals families were Dasypodidae (n ¼ 5) and Felidae (n ¼ 5).More than half large species (51%) were recorded exclusively in forest environments, 24.5% were recorded only in open areas and 24.5% in both environments.The species with greater number of records were Mymercophaga tridactyla (n ¼ 54) and Sylvilagus brasiliensis (n ¼ 30).The smallest number of records were obtained for Lycalopex vetulus (n ¼ 4) and Mazama americana (n ¼ 2).
The sampling effort of 18,144 hour*m 2 with mist nets in forest habitats resulted in capture of 107 individuals of 10 species of bats.The most frequently captured bat species were Carollia perspicillata (n ¼ 26) and Artibeus lituratus (19), and Mimon benetti was recorded only once.For small non-flying mammals, the capture success with live traps (Sherman and Tomahawk) was approximately 2%, with 20 species recorded.Pitfall traps did not resulted in exclusive additional records of species.However, two species of rodents (O.cleberi and P. roberti), and two species of marsupials (C.lanatus and M. demararae) were recorded exclusively with live traps.The Cricetidae family was the most representative, with 10 species, followed by Didelphidae (n ¼ 8).Five species were captured only in forest environments, three of them were restricted to open areas, and eleven were captured in both environments.Gracilinanus agilis was the most frequently captured small mammal (N ¼ 255), followed by Calomys tener (N ¼ 200).These two species were more abundant in dry season (C.tener -n ¼ 141 e G. agilis -n ¼ 209).Marsupials were reproductive during wet season, whereas rodents were reproductive during all year (Figure 3).About 24% of all G. agilis females captured on wet season were reproductive, whereas only 4% of them were reproductive during all dry season.The species C. tener was found reproductive throughout the year (Figure 3).The rodents Oecomys cleberi, Proechymys roberti, and Rhipidomys macrurus and the marsupials Caluromys lanatus and Thylamys karimii were the rarest small mammal species, with only one to three records each.Since field expeditions, we used this group to assess local populational and diversity parameters.

Discussion
We found mammal richness similar to those reported in long sampling period studies conducted in cerrado Protected Areas (Table 2).Despite our larger sampling effort, we recorded 73% of mammals recorded for Emas National Park,  ., 15(4): e0033, 2015 with about 132,000 ha (Rodrigues et al. 2002).A ´guas Emendadas Ecological Station, in the Federal District, with 11,000 ha, shows similar species richness (Marinho-Filho et al. 1998) to Serra do Faca ˜o region.Other studies found between 24 and 93 species for areas of different sizes and locations in the Brazilian cerrado (Mares et al. 1989, Schneider 2000, Moreira et al. 2008, Pereira & Geise 2009) (Table 2).
Small Mammals -We recorded 20 species of small nonflying mammals.About 25% of them were exclusive to forest environments and 15% were recorded only in open areas.This pattern is expected for the cerrado biome (Marinho-Filho et al. 2002).Considering studies in Protected Areas of cerrado and those that make use of pitfalls in addition to conventional traps, the richness of small non-flying mammals ranged from 19 to 29 species (Schneider 2000, Marinho-Filho et al. 2002, Pereira & Geise 2009, Carmignotto & Aires 2011, Bonvicino et al. 2012).This richness is similar (or even large) than our study, despite our larger sample effort (Table 2).The region of Serra do Faca ˜o has been greatly altered by human activity over the last 300 years, and now is also impacted by the fragmentation and reduction of natural habitats imposed by Sa ˜o Marcos river dams and by the infrastructure projects associated to Serra do Faca ˜o Hydroelectric Plant (Chaul 1997).However, the richness observed in the region of Serra do Faca ˜o is comparable to some Protected Areas in the cerrado domain, and other cerrado localities.We recorded some cerrado rare species (Marinho-Filho et al. 2002) as Oecomys cleberi, Micoureus demerarae, Caluromys lanatus, and Thylamys karimii, and their records were also rare throughout the study.The higher richness of forests when compared to open areas are also an expected finding for cerrado small non-flying mastofauna (Marinho-Filho et al. 2002).
The reservoir formation for the hydroelectric enterprise on Serra do Faca ˜o seemingly affected the small mammal diversity in open and forest habitats.The impacts of hydroelectric power plants on mammal communities were investigated by several other studies, and species loss is a common effect of reservoir filling (Cosson et al. 1999, Fournier-Chambrillon et al. 2000, Alho 2011, Andriolo et al. 2013).Besides habitat (and area) loss, the increase of predation and competition intensity are possible as factors affecting small non-flying mammals diversity loss in this kind of environmental change (Lemos de Sa ´1995, Alho 2011, Andriolo et al. 2013, Passamani & Cerboncini 2013).
Marsupials presented seasonal reproduction, whereas rodents were reproductive year round (Figure 3).The marsupial Gracilinanus agilis showed marked seasonal reproduction, confirming previous studies (Mares et al. 1989, Mares & Ernest 1995) but the rodent Calomys tener, was reproductive in both rainy and dry seasons (Figure 3).Studies carried out in central cerrado showed that the reproduction in C. tener females is significantly greater during the rainy season, with reproductive individuals recorded year round, but in lower proportions during the dry season (Mares et al. 1989, Rocha 2011).
Bats -Bat richness in Serra do Faca ˜o is lower when compared to other areas in cerrado, which presented a richness ranging from 16 to 25 species (Marinho-Filho et al. 1998, Aguiar 2000, Rodrigues et al. 2002, Bezerra & Marinho-Filho 2010).We recorded 10 species of bats, which represents only 10% of the richness known to the cerrado (Paglia et al. 2012).Although some other studies report equally modest numbers of bat species found in some cerrado areas (see Table 2 e.g.(Moreira et al. 2008) such low richness seems to be more an effect of relatively small sampling effort.The fact that we also only captured phyllostomid bats is expected with mist nets.The bat community from Serra do Faca ˜o is clearly undestimated and more species will be added with further sampling.However, there are some interesting records such as an individual of Mimon benetti a gleaning animalivore, not so common in inventories and collections.Although Anoura caudifer presents a wide distribution in South America,   occurring in several Brazilian states, there are few localities in the Brazilian cerrado with formal records in the literature (Oprea et al. 2009), and the species was recorded only in northeastern Goia ´s (Zorte ´a & Alho 2008, Bezerra & Marinho-Filho 2010, Peracchi et al. 2010).The present record of A. caudifer indicates this species presents a wider distribution across the cerrado.The small bat species richness verified at Serra do Faca ˜o may have been caused by a number of causes as well as their combination.1) We used only mist nets to sample bats and no bat detectors that help to find and identify species that fly high above the canopy of forests and are not easily captured with nets.2) Bat sampling was conducted in the period between  1997;1998-1999Schneider 2000 18:00 and 24:00 h, and consequently; 3) not all bat activity periods were covered and not all areas were explored as well as collection did not take place in all field trips.Thus, the assemblage of bats in Serra do Faca ˜o is clearly underestimated and further studies remain be done.Large mammals -We recorded 33 large mammals species (Table 1, Supplement 4), a richness higher than the sum of richness found in three Protected Areas in Brası ´lia (n ¼ 25) (Juarez 2008).Other studies also carried out in other cerrado regions, including Protected Areas (Marinho-Filho et al. 1998, Schneider 2000, Brito et al. 2001, Rodrigues et al. 2002, Moreira et al. 2008, Oliveira et al. 2009, Bocchiglieri et al. 2010, Carmignotto & Aires 2011, Alves et al. 2014), recorded 10 to 39 species of large mammals (Table 2).However, any comparison on large mammal richness between areas is limited because there is not a standard methodology used in different studies and the sampling efforts performed for this group in different studies are also very different.Indeed, most large mammals checklists are based on opportunistic data.
Eigth species of larger mammals surveyed are threatened with extinction (MMA 2014): Priodontes maximus, Mymercophaga tridactyla, Chrysocyon brachyurus, Lycalopex vetulus, Leopardus tigrinus, L. wiedii, Puma concolor, and P. yagouaroundi.The giant armadillo, Priodontes maximus, occurs in nearly all Brazilian biomes (Medri et al. 2010).This species, as well as their characteristic burrows, were found in both open and forest areas by direct observation and camera trapping.The fact that this species is considered extinct in various localities of southern Brazil (Marinho-Filho & Medri 2008) and is currently declining in other Brazilian regions, reinforces the importance of Serra do Faca ˜o for the conservation of its wild populations.Another threatened species found was the giant anteater, Mymercophaga tridactyla, which originally occurred in all Brazilian biomes, but is currently considered extinct in eastern states of Rio de Janeiro and Espı ´rito Santo, and its populations are declining in southern, southeastern and northeastern Brazil (Medri & Moura ˜o 2008).In the present study, the giant anteater was the most frequently recorded species among large mammals, using both open and forest areas -where most records were made.
The maned wolf, Chrysocyon brachyurus, a species associated with central Brazilian cerrados (Cheida & Santos 2010) was one of the rarest species in the present study, and the few records were obtained in open vegetation areas.Furthermore, we recorded the occurrence of some felines threatened with extinction.The margay, Leopardus wiedii, was spotted only in forest environments.Puma concolor, another feline threatened with extinction was the largest predator, with few photo records in the region.We did not detect jaguar, Panthera onca, which could explain the considerable abundance of mesopredators, such as L. wiedii in the area.The oncilla, Leopardus tigrinus, was one of the rarest species among larger mammals and this may be explained by the relatively high frequency of L. pardalis, one of the largest predators found, and capable of excluding smaller spotted cats in areas where it is dominant (Oliveira 2004).
Despite the fact that Serra do Faca ˜o region has been altered for centuries by several kinds of natural resources exploitation that resulted in a quite fragmented landscape, and the area that we sampled is not included in or near to any protected area, the present study reveals that it still shelters a mammalian fauna corresponding to at least 25% of the total mammalian fauna of the whole cerrado biome.This richness, in addition to the presence of rare and threatened species, reinforces the importance of natural remnants of cerrado for the conservation of regional mammalian fauna.It also calls attention for the importance of enforcing the protection of natural areas that may be affected by many large infrastructure projects.

Figure 1 .
Figure 1.Serra do Faca ˜o state of Goia ´s, and location of sample sites a) Brazil; b) River system in the area; c) Sample Sites.O -Open areas; F -Forest.Description of sampling points and coordinates are on Supplement 1.

Figure 2 .
Figure 2. Species accumulation curve (collector's curve) based on effort for all mammalian species recorded in the Serra do Faca ˜o hydroelectric plant from May 2008 to September 2013.

Figure 3 .
Figure 3. Percentage of reproductive females of marsupials and rodents captured during dry and wet seasons at Serra do Faca ˜o region, Goia ´s state, from May 2008 to September 2013.

Figure 4 .
Figure 4. Rarefaction curves (Cole estimator) for small non-volant mammals captured in open and forest environments at Serra do Faca ˜o, Goia ´s state, from May 2008 to September 2013.

Figure 5 .
Figure 5. Rarefaction curves for small non-volant mammals in open areas before (May 2008 to November 2009) and after (February 2010 to September 2013) the reservoir flooding of the Hydroeletric Power Plant of Serra do Faca ˜o.

Figure 6 .
Figure 6.Rarefaction curves for small non-volant mammals in forest enviroments before (May 2008 to November 2009) and after (February 2010 to September 2013) the reservoir flooding of the Hydroeletric Power Plant of Serra do Faca ˜o.

Table 2 .
Mammalian richness from different Cerrado localities, including Protected Areas (PA).