Bat ( Chiroptera ) assemblages in three Cerrado fragments of Mato Grosso do Sul , southwestern Brazil

Despite its high species richness and large area, the Brazilian Cerrado is a generally undervalued and underprotected biome. There are relatively few published studies of bat assemblages in this area. We surveyed for bats using mistnets from April–November 2013 within and around Campo Grande, Brazil in an urban fragment, an agricultural fragment, and a larger fragment of continuous forest. We captured 508 individuals from 21 species representing four families: Phyllostomidae (10), Molossidae (6), Vespertilionidae (4), and Noctilionidae (1). Phyllostomids accounted for 91.73% of captures. The most common species were Artibeus planirostris (27.76%), Artibeus lituratus (21.06%), and Sturnira lilium (11.61%). There was variation between the sites: richness and diversity was highest in the continuous forest and lowest in the urban fragment. Evenness was highest in the rural fragment. The least similar sites were the urban fragment and the continuous forest; similarity was greatest between the rural site and continuous forest. DOI: 10.15560/10.6.1380 L s t s o f s p e c ie s


Introduction
Brazil's Cerrado is one of the most biodiverse savannas in the world, with high levels of endemism (Mendonça et al. 1998;Ratter et al. 2003), and is considered a biodiversity hotspot (Myers et al. 2000). At over 2 million km 2 , it is the second largest biome in the country, covering almost 24% of the national territory (IBGE 2004). Despite its ecological value, it is under-protected compared to other biomes in Brazil (Ratter et al. 1997;Silva and Bates 2002) with less than 3% of its area legally protected and almost 55% of its original extent deforested or degraded (Machado et al. 2004).
Brazil is home to 174 species of bats (Paglia et al. 2012), 103 of which can be found in the Cerrado (Aguiar and Zortéa 2008). However, according to Bernard et al. (2011), only 6% of this biome has been minimally surveyed. Further, the studies within the Cerrado are not welldistributed throughout its full area (Aguiar and Zortéa 2008). In Mato Grosso do Sul in particular, studies tend to focus on the Pantanal, leaving the Cerrado understudied. Therefore, there is far more data on bats of the Pantanal than the Cerrado in this state (Cáceres et al. 2008).
Mato Grosso do Sul has 65 species of bats currently recorded, 49 of which can be found in the Cerrado (Cáceres et al. 2008;Bordignon and Santos 2010;Bordignon et al. 2011;Santos and Bordignon 2011;Silveira et al. 2011). In Campo Grande, there are records for 24 bat species, which comprise 23% of Cerrado species and 39% of bat species found in the state of Mato Grosso do Sul (Pulchério-Leite et al. 1999;Ferreira et al. 2010).
Our objective was to evaluate the species richness, abundance, and biodiversity of bats in three Cerrado fragments: one within an urban matrix in Campo Grande, one within a rural, agricultural matrix on the edge of the municipality, and a third larger fragment of continuous Bat (Chiroptera) assemblages in three Cerrado fragments of Mato Grosso do Sul, southwestern Brazil production. Most remaining vegetation is typical of the Cerradão. It also includes patches of Cerrado sensu stricto, and veredas, humid areas around small lakes (Ratter et al. 1997;Silva and Bates 2002). These areas of natural vegetation are small and surrounded by agricultural fields and pasture. Although the surrounding landscape is mostly agricultural, urbanization is increasing. For example, during the time of the study, the main road that turns off from the city to this rural zone was paved.
The third site is Fazenda Piana (FPia), a private reserve of approximately 6.5 km² connected to forested hills with an elevation of 390-490 m above sea level that extend up to 13 km from the site. The area is used for both agricultural and touristic activities; the larger landscape surrounding Fazenda Piana is primarily composed of large monoculture fields. It is located in Sidrolândia, a rural municipality about 40 km southwest of Campo Grande. 60% of FPia is covered by native Cerrado vegetation, ranging from shrubs to semideciduous forest (Ratter et al. 1997;Silva and Bates 2002). We sampled between the coordinates 20°48′19″ S, 54°50′36″ W and 20°47′46″ S, 54°50′42″ W. The region is part of the geological formation of the Serra de Maracaju, a chain of hills that crosses the central portion of Mato Grosso do Sul.

Data collection
Surveys were conducted twice per month from April-May and July-November 2012 at each site in the week leading up to the new moon, to reduce the interference of moonlight (Morrison 1978;Lang et al. 2006). Nets were opened around sunset and extended for six hours (Kunz and Kurta 1988;Bergallo et al. 2003). We used five mistnets each night, totaling 1,836 m 2 ·h mo -1 of sampling effort in each site. Total sampling effort for all sites was 38,556 m 2 ·h (Straube and Bianconi 2002). Mist nets were placed in a variety of available habitats within the sites (e.g., open areas, areas of dense shrubby vegetation, within forested areas, along forest edges, near streams or ponds), and close to possible roost sites. Each night, nets were placed in a different location within each site (Kunz and Kurta 1988;Bergallo et al. 2003).
Bats were removed from the mist-nets and placed in cloth holding bags. Each bat was identified by species, using external morphology, measurements, and dentition, according to Vizotto and Taddei (1973), Charles-Dominique et al. (2001), Gregorin and Taddei (2002), Reis et al. (2007) and Gardner (2008). One specimen of each species from each area was collected and deposited in the zoology collections at the Federal University of Mato Grosso do Sul as a voucher specimen following university guidelines (Appendix 1). Specimens which could not be identified in the field were also euthanized and brought to the lab for further identification. All other bats were released.

Data analysis
Calculations of the Shannon-Wiener index H′, evenness and dominance were conducted using the abundance of each species for each site in PAST (Hammer et al. 2001). Similarity between sites was compared using the Jaccard coefficient, calculated by hand, and the Shannon Diversity t-Test, using PAST (Hammer et al. 2001). We also assessed the difference in species richness and composition by feeding guild using a t-test based on species richness per guild in PAST (Hammer et al. 2001).
Rarefaction curves were calculated using PAST in order to estimate the completeness of our survey (Hammer et al. 2001). Estimates of total species richness per site and overall were calculated using the Chao 2 index classic formula with 100 randomizations in EstimateS Version 8.2.0 (Colwell 2005).
Species composition and richness varied between the sites. Nine species were captured in FUrb. The Shannon-Wiener diversity in this site (H′ = 1.68) was the lowest among the three, while FPia presented the highest diversity (H′ = 2.10) with 18 species captured. In FRur, 10 species were captured (Fig. 2). The value of the Shannon-Wiener index of FRur (H′ = 1.95) was closer to FPia than to FUrb.
It is also possible that the low evenness at FPia could be partially attributed to the matrix around it. Although the chain of forested hills extends for up to 13 km around FPia, it is surrounded by a matrix of large monocultures. Large extents of homogenous monoculture can negatively affect bat species richness and diversity, even in habitat patches within these matrices (Gehrt and Chelsvig 2003;Numa et al. 2005).
Divided by trophic guilds, we found 10 species of insectivores, eight species of frugivores, one species of nectivore, one species of sanguivore, and one species of piscivore ( Figure 3). Despite high species richness, the number of individual insectivores captured was low, while frugivores had relatively high species richness as well as a large number of individuals. There was no significant difference in species richness per guild between any of the sites (FPia and FUrb: t = 2.449, p = 0.07048; FPia and FRur: t = 2.359, p = 0.07774; FUrb and FRur: t = 1, p = 0.3739).

Number of Species
Frugivore Nectivore Insectivore Sanguivore Piscivore lower quality habitat, even at FPia (Fenton et al. 1992;Medellín et al. 2000). However this could also indicate the need for further sampling at this site, as the rarefaction curves have not yet reached an asymptote (Figure 4) (Moreno and Halffter 2000). Two molossid species, Nyctinomops laticaudatus (E. Geoffrey, 1805) and Molossus molossus (Pallas, 1766), were found only in the urban site. Molossid bats are known to adapt well to urban environments due to the availability of insect prey attracted by city lights (Bredt and Uieda 1996;Avila-Flores and Fenton 2005;Reis et al. 2006;Lima 2008) and their ability to use artificial roosting sites such as crevices in buildings (Romano et al. 1999;Perini et al. 2003;Mendes et al. 2011). However, due to the difficulty of capturing insectivorous species with mist-nets (Pedro and Taddei 1997;Bergallo et al. 2003;Cunto and Bernard 2012), we cannot be certain of the absence of either N. laticaudatus or M. molossus at FRur or FPia.
Eight species were captured only at FPia, including two dietary specialists-Desmodus rotundus (E. Geoffrey, 1810) (sanguivore) and Noctilio leporinus (Linnaeus, 1758) (piscivore). The abundance and availability of food sources, e.g., livestock for D. rotundus (Delpietro et al. 1992;Numa et al. 2005) and artificial lakes stocked with fish for N. leporinus (Zortéa and Aguiar 2001) at FPia probably contribute to the presence of these two species at this site. The large proportion of species found only at FPia (38.09%), as well as the high level of diversity (H′= 2.10), point to the importance of maintaining patches of natural habitat for the conservation of bats in the Cerrado.
The rarefaction curves for FUrb and FRur were reaching their asymptote, while the rarefaction curve was still rising steeply at FPia (Figure 4). Using nights of capture as the unit of sampling effort, the species richness estimated by the Chao 2 index was 24 (± 1.36) across all three sites, indicating the capture of approximately 87.5% of estimated species richness. The estimated species richness for FUrb (12 ± 1.69, 75%) and FRur (12 ± 1.25, 83.33%) were similar while the value for FPia was higher (21 ± 1.55, 85.71%) ( Figure 5). Continued research, especially the use of acoustic monitoring and active roost searches, in addition to mist-netting, could increase the number of species recorded in all three sites (Cunto and Bernard 2012) and further contribute to our knowledge of the bat communities of the Cerrado, which is generally lacking (Bernard et al. 2010).
Although a lack of replicates for each habitat type prevents us from drawing general conclusions on the effect of land-use change on the bat community of the Cerrado, our results indicate that conversion of savanna habitat to urban or agricultural areas may have negative effects on bat species richness and diversity. Bats provide valuable ecosystem services: controlling insect populations, including agricultural pests, dispersing seeds, and pollinating plants, a number of which are of economic value to humans (Kalka et al. 2008;Williams-Guillén et al. 2008;Boyles et al. 2011;Kunz et al. 2011). In the Cerrado in particular, bats have been shown to be important pollinators, carrying pollen long distances between isolated tree stands (Moraes and Sebbenn, 2010). Further, by dispersing seeds, especially those of important pioneer species, phyllostomid frugivores play a role in the secondary succession and the regeneration of degraded forests (Gorresen and Willig 2004; Kelm et al. 2008;Kunz et al. 2011). Given these ecological functions and their importance for human activity, declines in bat populations and species richness are cause for concern. Further research on the effects of landscape changes on bats in the Cerrado is needed.