1. Studied Area.

The cave studied is the "Gruta do Riacho Subterrâneo" (23°16'10.40"S, 47°13'49.79"W, 726 m above sea level), located in the city of Itu - state of São Paulo, Brazil, and belongs to Itu's post-orogenic granite suite (Grupo Pierre Martin de Espeleologia 2014) (Figure 1 A, B). This cave is amongst the six largest granite caves in the world, and it is so far the largest of the Southern Hemisphere, with approximately 1,415 meters of linear development, formed by the superposition of several blocks. The cave has many entrances offering various routes for bats.

Source: ^{Grupo Pierre Martin de Espeleologia, 2014}

Figure 1 (A) Maps showing the location of Gruta do Riacho Subterrâneo in South America and in Itu municipality, SP. (B) Geological formation 

The region where the cave is located is characterized by humid subtropical climate with dry winters and warm summers (Cfa) according to Köeppen (1948). The regional vegetation is a transition of Cerrado, ombrophilous dense forest, stational semidecidual forest, and the ecotone zone among these vegetations (^{Rizzini 1997}). Vegetation around the cave is nevertheless under strong anthropogenic pressure since the beginning of economic development in the São Paulo state highlands (Grupo Pierre Martin de Espeleologia 2014). In the surrounding area there is a small stream close to the cave and another in its interior, providing water resources to the bat assemblage and other communities present. Also, there is a small cave at a few meters distance and a large one at approximately 6 Km. It must be pointed out that several fires happened in 2010 in the surroundings of Gruta do Riacho Subterrâneo cave (Bichuette, M. E. personal observation) evidencing another type of disturbance the cave and its surrounding area undergo, making them even more altered.

Due to the granitic origin of the cave it is difficult to precisely establish the amount of existing saloons, since this is a non-regular cave along its whole extension being formed by superimposed granitic blocks, thus offering many entrances and possible routes to bats.

2. Bat Samplings and Environmental Variables Measurements

Samplings were carried out every two or three months in the period from October 2013 to September 2014. The first sampling (pilot) took place in August 2013, in order to define where the mist-nets could be placed. Table 1 shows the sampling dates, with October, December and March representing the rainy season and June and September representing the dry one (^{INMET 2014}).

Four or five mist-nets were set at the entrances of the caves in each of the three sampling days. Two other nets were set outside the cave (in a pasture and near a stream), in the first and in the third sampling days. These places were defined considering their dimensions and accessibility (Figure 2). It is important to emphasize that the cave has many entrances, so it is difficult to cover all possible routes for bats. The mist nets had three different sizes (4 x 3 m; 6 x 3 m and 7 x 3 m), installed in according to the entrance sizes. In the pasture we used a 7 x 3 m net and close the stream a 4 x 3 m net. The nets were set before sunset and were exposed for five hours, with regular checks every 20-30 minutes.

Author: Ives Arnone

Figure 2 A view of sampling mist-nets at Gruta do Riacho Subterrâneo cave, Itu municipality, SP. 

After captured, the specimens were identified on-site at species or genus level, and some were kept to allow a more accurate identification in the laboratory by utilizing specific identification keys (^{Vizzoto & Taddei 1973}, ^{Gardner 2007} and ^{Reis et al. 2013}). The specimens were deposited in the reference collection at Subterranean Studies Laboratory of the Federal University of São Carlos (LES - UFSCar).

Measurements of temperature (°C) and relative humidity (%) were obtained with a thermo-hygrometer (Instrutherm THAL-300; 0.1 resolution ± 5.0% accuracy) and monthly precipitation values were obtained from the National Institute of Meteorology (INMET, 2014), from the nearest station of Itu municipality. The moon phase was recorded in each collection day and posteriorly checked in a moon phase Table for each sampling day and month. To attribute a value for the moonlight luminosity of each lunar phase we used the following scale: new moon = one, horning = two, waning = tree, and full moon = four, as usually suggested for quantitative analysis based on qualitative evaluations.

3. Data Analysis

Capture effort was calculated according to ^{Straube & Bianconi (2002)}. A sample-based species accumulation curve was plotted to analyze the Sufficiency Sampling Size, using "EstimateS" software (^{Colwell 2005}). The relative abundance of each species was calculated and expressed as a percentage of the total assemblage abundance, in order to show their representativeness in the assemblage.

A multiple regression analysis was performed to verify the influence of the abiotic variables (temperature, relative humidity and pluviosity) over the abundance of the assemblage. Later on, the Akaike information criterion (AIC) was applied, in order to select the best representative model (Akaike 1970). A canonical correspondence analysis (CCA) was performed to verify the influence of the abiotic data matrix (temperature, relative humidity, moonligth and pluviosity) over the species abundance and assemblage richness. These analyses were carried out using the statistical software "R Development Core Team 2011".

1. Bat Assemblage Composition

Among the 15 species recorded in this study only eight were found in the interior of Gruta do Riacho Subterrâneo, whereas the other seven were captured in the surroundings. From the eight species found in the interior of this granitic cave, six were already recorded by ^{Trajano (1985)} and by ^{Arnone (2008)} in Vale do Ribeira Atlantic Rainforest caves. Individuals from two genera, Myotis sp. and Micronycteris sp. could not be included in the comparison because they were not yet identified to the species level.

Among the seven species collected only in the surroundings of the Gruta do Riacho Subterrâneo (pasture or native vegetation) six have also been previously recorded by ^{Trajano (1985)} and by ^{Arnone (2008)} in Vale do Ribeira caves. Only one species, Epitesicus diminutus, captured in the pasture nearby the cave, was not previously recorded by these two authors in Vale do Ribeira caves. According to ^{Reis et al. (2013)} this species is found in primary and secondary forests and forest fragments, but also in open areas and urban environments. In São Paulo State it was previously recorded in the Atlantic Forest at Morro do Diabo State Park, where it is also a rare species (^{Reis et al. 1995}).

Only three species abundant in the assemblage studied, the species C. perspicillata and D. rotundus and Myotis sp.. Two of them were also abundant in bat assemblages of other caves in São Paulo state (^{Trajano 1985}, ^{Campanhã & Fowler 1993}, ^{Arnone 2008}).

Comparing the species composition of the bat assemblage in the granitic cave Gruta do Riacho Subterrâneo with those from caves with different lithologies in São Paulo State, as the calcareous caves of Alto do Ribeira (^{Trajano 1985}, ^{Arnone 2008}) and the sandstone caves in the Preservation Area of Corumbataí (^{Campanhã & Fowler, 1999}) it was found that they have very similar composition, although differing in the number of species. The assemblages of 36 caves so far inventoried probably belong to the same regional chiropterofauna, thus suggesting that cave lithology is not a determinant factor for bat assemblages taxonomic structure. One possible explanation for this similarity and also for the relatively high number of bat individuals captured in Gruta do Riacho Suberrâneo could be the fact that all these caves are located in remnants of the Atlantic Rainforest.

Bats from Gruta do Riacho Subterrâneo and its surroundings were classified into five trophic guilds with predominance of the frugivore guild, a fact that is corroborated in other studies that indicate that there is a predominance of this guild in the Neotropical Region (^{Sipinski & Reis 1995}, ^{Ribeiro-Mello 2009}, ^{Stevens & Amarilha-Stevens 2012}, ^{Bobrowiec et al. 2014}, ^{Lourenço et al. 2014}). The predominance of frugivorous could, at least in part, be influenced by the sampling method used (mist nets) which is more effective for this guild, but less effective for insectivores due to their more refined echolocation system (^{Handley 1967}, ^{Trajano 1985}, ^{Arita 1993}, ^{Sipinski & Reis 1995}).

Some species were rare being captured only once (Anoura geoffroyi, Micronycteris sp., Chiroderma doriae, Sturnira tildae and Eptesicus diminutus), which could be due to the rarity of the species themselves, which usually tend to form small colonies and also to travel long distances to obtain food (^{Stevens & Amarilla-Stevens 2011}). For instance, Micronycteris sp. live in small groups of six or less individuals (^{Emmons, 1990}). Chiroderma doriae has not been found in the caves in the studies of ^{Trajano (1985)} and of ^{Campanhã and Fowler (1999)}, but it was recorded by ^{Arnone (2008)} in Vale do Ribeira caves, who captured only one individual in this study site. It also occurred in small numbers in a semi-deciduous forest in the Ecological Station of Caetetus, São Paulo state (^{Pedro et al. 2001}).

2. Seasonality

The results of multiple linear regression analysis did not show significance for the correlations among abiotic variables and the abundance of bats captured along the year. On contrary, the multivariate canonical analysis showed a significant positive relationship between the relative humidity and bat abundances. In the case of multiple regression, the tested variables are redundant (relative humidty is influenced by pluviosity and air temperature), which probably caused a noise in the analysis. Certainly the pluviosity and air temperature (affecting the relative humidity) influence the bat abundances and the canonical analysis supported this.

Corroborating this idea, we observed that the lowest captures of bats occurred in October 2013 and September 2014, which correspond to the transition period between dry and rainy seasons at the studied site. Bat populations were affected by seasonality, since the greatest bat abundances occurred in the warm and rainy summer months: December 2013 and March 2014, although a high number of bats was also captured in June 2014. The population of Carollia perspicillata, the most abundant species in Gruta do Riacho Subterrâneo cave followed this pattern, also displayed by less abundant species, such as Sturnira lilium and Glossophaga soricina. These species belongs to the frugivorous and nectarivorous trophic guilds, that are directly dependent on flowers and fruit production, which in Atlantic Rainforest and Cerrado have pronounced seasonality (^{Morellato et al. 2000}; ^{Batalha et al. 2004}; ^{Marques et al. 2007}).

The different pattern observed for Desmodus rotundus, which showed high abundance along all the year (except for the last capture event in September 2014) is probably related to its hematophagous trophic guild, dependent only on the availability of mammals blood (as cattle and horses), its main food source, which is constant throughout the year in the region.

^{Stevens & Amarilla-Stevens (2011)} in a study on bats in the Atlantic Forest in Paraguay found that some species displayed greater population changes in abundance, which were more related with seasonality than others. For example, the species Artibeus lituratus had a decline of more than 30% of its population during the colder season, but other species such as Carollia perspicillata and Artibeus fimbriatus did not show significant population density changes. The pattern observed for the latter species is similar to that of Carollia perspicilata, in which there was no direct relationship between this species abundance and temperature changes. The species A. lituratus was also captured, but its extremely low abundance in Gruta do Riacho Subterrâneo, with only two captures during all study, does not allow inferences regarding seasonality. Also, ^{Ribeiro-Mello (2009)} and ^{Ortencio-Filho et al. (2010)} showed that changes in temperature throughout the year influences the structure of bat community, and that this can be associated with a change in the feeding sites due to the decrease in food availability in the local area. Similarly, in Gruta do Riacho Subterrâneo cave, the lowest abundances of bats in September 2014 were probably due to the lower food availability, since this was the driest period in this one year sampling.

Another environmental factor that can influence cave bat capture is the moonlight intensity. It was shown by ^{Crespo et al. (1972)} in Mexico that Desmodus rotundus, a haematophagous species, and by ^{Morrison (1978}, ¹⁹⁸⁰⁾ that some frugivorous species in Panamá, Canal Zone, presented lunar phobia. In the present study there was no relationship between the number of bat captures and the moon light phases.