The study area is in the region known as Lagoa Santa Karst in the municipality of Matozinhos, state of Minas Gerais, Brazil (Fig. 1). The vegetation in the region is transitional between the Atlantic Forest and Cerrado (Brazilian savanna) biomes. The region has more than 500 recorded caves (Hermann et al., 1998), which provide roost for several species of bats (Torquetti et al., 2017). The region has a seasonal climate, with a rainy season from October to March and a dry season from April to September, and an average annual temperature of around 21 °C (Hermann et al., 1998).

Fig. 1.

Study area showing sites where acoustic data were obtained. Circles indicate sampling sites in pasture, triangles in forest and squares in eucalyptus plantation.

(0.46MB).

Encompassing about 1800 ha, the study area is characterized by an agropastoral landscape matrix with approximately 40% of its area covered by fragmented native forest, mainly on karst rock outcrops, about 40% used for pasture of cattle and horses and 20% covered by monocultures of eucalyptus forests. We collected data on three habitat types, namely semideciduous forest, pasture and Eucalyptus globulus monocultures, hereinafter referred to as forest, pasture and eucalyptus, respectively.

We chose five sites in each habitat type (15 total) for bat acoustic sampling (Fig. 1). The sites were defined with a minimum distance of 1 km between them and a minimum distance of 250 m to the edge of another habitat type, with the aim of obtaining independence of sample sites. We used two autonomous Song Meter 2 Bat + ultrasonic recorders (Wildlife Acoustics, Concord, USA), which were field mounted at 1.5 m above ground level and with the microphone tilted at 45 ° to the sky. We scheduled the recorders to activate at twilight hours (18:00 h) and to remain active until 06:00 h the following day.

Acoustic data were collected in the months of November 2019, January 2020 and February 2020, with 15 sampling days per month. Each night, we randomly chose two sites belonging to the same habitat type to be sampled and 1 recorder was installed at each. Thus, each site was sampled once a month, totaling 15 samples for each habitat type and a total of 45 samples. We chose the rainy season for sampling because, in the Neotropics, this season presents greater availability of food resources, which is reflected in greater bat activity (Fenton and Zortea, 1999). In addition, collections took place during the dark phase of the moon, as there is evidence of greater bat activity on these occasions (Saldaña-Vázquez and Munguía-Rosas, 2013).

The independence among sampling sites was investigated using the Mantel Test, to determine whether the sites can be treated as true replicas. This test has been performed using species/sonotypes passes to calculate dissimilarity matrices with the Bray-Curtis Index, with 9999 randomizations.

Spectrograms and acoustic identification of bat sonotypes were performed using RAVEN software (Cornell Lab of Ornithology Bioacoustics Research Program). A bat pass was considered as the sequence of two or more echolocation calls (or pulses) recorded the moment after the recording was triggered (Schnitzler and Kalko, 2001). To avoid recording multiple passes of the same individual, we considered an interval of two minutes before recording another pass of the same sonotypes. Feeding buzzes and social calls were counted with no break between passes. We visually inspected the spectrograms of all recordings to identify each recording to species level based on the following acoustic parameters: structure of the echolocation pulse, highest emitted frequency, lowest emitted frequency, peak frequency, pulse duration and interval between pulses. We classified sound recordings as sonotypes when identification to species level was not possible.

We characterized forest vegetation structure and food availability in each habitat type to see if these characteristics could influence the responses of aerial insectivorous bats.

We extracted 2569 general bat passes, 502 feeding buzzes and 43 social calls of insectivorous bats for analysis. We registered 1477 passes in pasture, 955 in forest and 137 in eucalyptus (Table 1). We recorded 326 feeding buzzes in pasture, 170 in forest and only six in eucalyptus. On the other hand, we recorded 34 social calls in forest, followed by eight in pasture and only one in eucalyptus. It was possible to identify seven species and four sonotype groups (Table 2) belonging to the families Vespertilionidae (Fig. S2 in Supporting Information) and Emballonuridae (Fig. S3 in Supporting Information). The most frequent taxa were Myotis sp.1 (571 passes), Histiotus velatus (443 passes), Peropteryx sp. (423 passes) and Eptesicus furinalis (335 passes) (Table 2); Myotis ruber had the lowest number of passes (8) (see Table 2). The Mantel Test showed that the distance between sample sites did not affect bat composition, thus indicating independence among sites (R = −0.065, P = 0.745).

The activity of the insectivorous bat assemblages differed among habitat types. The PERMANOVA analysis (P = 0.006; Table S1 in Supporting Information) and the distribution of points in the NMDS plot, suggests that the bat species/sonotypes activity is different in relation to the habitats (stress = 0.100; Fig. S5 and Table S2 in Supporting Information). Species/sonotypes composition did not differ between forest and pasture, however, the composition of these two habitat types differed significantly from that of eucalyptus (PERMANOVA, P = 0.237, Table S1 in Supporting Information). This result is also observed through the NMDS plot (NMDS stress = 0.091; Fig. S6 and Table S3 in Supporting information).

Pearson correlations showed collinearity between feeding buzzes and the covariate general passes (R = 0.943, P = 0.007), and so we excluded feeding buzzes as a response variable in subsequent GLM analyses. The variables general passes and species/sonotype richness did not differ significantly between forest and pasture, however, both variables in these two environments differed significantly from the values found in eucalyptus (Table 1; Fig. 2a). No significant differences were detected for social calls (GLM: F = 2.248, P = 0.100; Table 1).

Fig. 2.

a) Boxplot comparing bat activity and species/sonotypes richness in each habitat type. Black dots represent sampling sites. Points outside the whiskers are outliers. Dark boxes indicate significant differences by contrast test. b) Linear regression of bat activity and species/sonotypes richness as a function of canopy density. All responses were canopy density dependent.

(0.29MB).

The explanatory variable canopy density had a negative effect on the general activity and species/sonotype richness of bats (GLM: F = 6.818, P = 0.009; Table 1; Fig. 2b). Finally, the variables understory density and food availability had no effect on bat activity (GLM: P > 0.05, Table 1).

The composition of insectivorous bat assemblages found in forest and grassland habitats is characterized by the presence of species/sonotypes belonging to the guilds of open space foragers (emballonurid species) and edge space foragers (vespertilionid species) (Denzinger and Schnitzler, 2013), which possess morphological, sensory and motor characteristics that favor foraging at the forest edge, close to the canopy line, between treetops or in open spaces, avoiding the interior of forests (Denzinger et al., 2016).

The similarity among assemblages indicates that different resources available in these habitats, such as food and roosts (Mendes et al., 2016; Monck-Whipp et al., 2018), may be influencing their use by bats, as many roosts are available on rocky outcrops within these forests and are used by several of these species, as verified by previous studies (Talamoni et al., 2013; Torquetti et al., 2017). Considering that insect availability does not have a significantly different effect among habitats, these aerial insectivorous bats probably move between forest and pasture in search of satisfying their foraging and roosting needs. The bat activity observed in pasture, and considering its correlation with feeding buzzes, may indicate that species benefit from these open areas by using them as foraging sites. Thus, apparently the resources present in forest and in pasture allow these insectivorous bats to use these areas in a complementary way (Dunning et al., 1992).

The bat activity observed in pasture demonstrates that, apparently, in the studied landscape, the conformation of pastures interspersed with areas of native habitat influences insectivorous bat activity, which may reduce the distance between roosting and foraging sites. A similar result was found in Chile, where bat activity levels can be significantly associated with the amount of forest in relation to their fragmented surroundings (Rodríguez-San Pedro and Simonetti, 2015). Furthermore, it has been demonstrated that bat activity in areas of extensive grazing is greater at the edges than in the interior (Rainho and Palmeirim, 2011), since within vast pasture areas bats become more exposed to adverse weather conditions and predators.

The results regarding eucalyptus plantations demonstrate that they have a negative effect on bat activity and species/sonotypes richness, as also observed for phyllostomid bats in an area of Brazilian Cerrado (Pina et al., 2013), although the cause is not clear. Some studies carried out in subtemperate regions in the Neotropics, where native forests have been replaced by exotic forest plantations, show that some bats species are able to use these areas to move, forage and roost (Barros et al., 2014; Rodríguez-San Pedro and Simonetti, 2015).

A hypothesis that can be raised to explain our results is related to the canopy structure of the studied stands. The plots are recent and were planted where there were other pastures. The trees were planted relatively close together and their growing branches are intertwined, which can make it difficult for bats to navigate inside these plantations, since the wing morphology of the recorded species is not suitable for highly maneuverable flights, as they are adapted for flying in open areas (Schnitzler and Kalko, 2001). Furthermore, the absence of understory may have contributed to this result. The occurrence of understory with a complex vegetation structure and with a height above 3 m, common in mature eucalyptus plantations, is associated with higher levels of bat activity and species richness (Simonetti et al., 2013; Cruz et al., 2015; Burgar et al., 2017).

The finding that activity and species/sonotype richness levels decrease as canopy density increases can be expected, since the characteristics of the studied landscape apparently favor the permanence of these open and edge space foragers. The lack of a relationship between bat activity and food availability must be related to the fact that we did not find significant differences in insect abundance among habitats. However, a study carried out in Panama (Estrada-Villegas et al., 2012) also did not observe an effect of insect availability on insectivorous bat activity.

The experimental design of this study has some limitations, such as the small number of sampling sites, due to the small size of the study area. Although this, we consider our results important because our data reinforce the importance of maintaining preserved areas of native vegetation in agropastoral landscapes and demonstrates that exotic forest plantations can exert a negative effect on insectivorous bats assemblage.