INTRODUCTION

The availability of food and roosts are key factors that limit the distribution, population density and reproductive biology of bat species (^{Dechmann & Kerth 2008}). The importance of diurnal roosts for bats has been long known (^{Kunz 1982}), but the evolutionary mechanisms underlying the interactions between bats and their roosts are still poorly studied. Recent evidences from fruit-eating bats, showed correlations between the use of foliage and leaf tents as roosts, with morphological and ecological attributes, considering pelage patterns, group size and social organization (^{Santana et al. 2011}; ^{Garbino & Tavares 2018}; ^{Tavares et al. 2018}).

Most bats may compete for the use of roosts and many species do not actively build them (^{Kalko et al. 2006}). There are, currently, more than 1386 bat species known (^{Burgin et al. 2018}) and only approximately 35 species use structures modified by themselves to roosting, such as leaves, stems, parts of plants (^{Kunz & McCracken 1996}; ^{Kunz & Lumsden 2003}), and ant and termite nests (^{Dechmann et al. 2009}; ^{Chaverri & Kunz 2010}). Both modification and use of termite nests is rare among bats, and only three species has been documented to have this roosting habits: Lophostoma silvicolum^{d’Orbigny, 1836}, Lophostoma carrikeri (^{Allen 1910}) and Lophostoma brasiliense^{Peters, 1867} (^{York et al. 2008}; ^{Voss et al. 2016}).

It is known that males of L. silvicolum use their teeth to modify active termite nests and use them as roost, suggesting adaptations to roost excavation. Besides, these adaptations could represent advantages for reproductive success of the male, because females use termitaries as maternity roosts (^{Dechmann et al. 2009}). Thus, the ability of males to modify and use termite nest appears to be a behavior selected by females (^{Dechmann & Kerth 2008}). Although such interactions with termite nests have been explored in L. silvicolum, the roosting ecology is poorly understood for other species of Lophostoma.

There are reports of roosting L. brasiliense, in termite nests, from several sites along its neotrop ical distribution, including Belize (^{Reid 1997}), Brazil (^{Peracchi & Albuquerque 1993}), Costa Rica (^{York et al. 2008}), Panama (^{Handley 1966}), Peru (^{Kalko et al. 2006}; ^{Voss et al. 2016}), Trinidad and Tobago (^{Goodwin & Greenhall 1961}), and Venezuela (^{Robinson & Lyon 1901}; ^{Handley 1976}). Nevertheless, they are based on occasional encounters and do not provide details about social organization, architecture and size of termite nests that are used by the bats, which allows us to see possible patterns in their choice as shelter. Also, there is not a comprehensive compilation of records of bats in termite nests to serve as a base to analyzing variations on the roosting behavior of termite nest’s bats.

We herein report observations on the use of termite nests by Lophostoma brasiliense in Colombia, including information on both diet and activity patterns. Also, we conduct a comprehensive compila tion of records on associations between termite nest and bats, in order to investigate the frequency of use of this shelter type by Chiroptera.

MATERIAL AND METHODS

Study area

The study was conducted at Santa Librada Agroecological Reserve (SLAR), in El Líbano municipality, on department of Tolima over the central cordillera of Colombia, at 800 m elevation (4.8758 N; 75.0225 W; Fig. 1). The area encompasses 52 ha of secondary forest in distinct stages of succession, mixed in a mosaic of coffee and cacao plantations, as well as native species, such as Anacardium excelsum, Annona cherimola, Carludovica palmata, Cecropia peltata, Cedrela montana, Erythrina fusca, Heliconia bihai, Lafoensia acuminata and Ochroma pyramidale. The average annual temperature is 23°C and the average annual rainfall ranged from 1500 to 2000 mm.

Roost and group/colony characterization

We made one observation per each shelter of L. brasiliense (two termite nests and one underneath houses basement), in January 2016 and January 2018, for a total of six observations. We recorded width, length, height, and diameter measurements for each of the termite nest, using a one-meter professional measuring rule. To count the number of individuals, present in each shelter and in every visit, we capture the individuals using an entomological net and mist-nets placed around the nests.

Each captured individual was sexed, identified, measured with a 0.01 mm-accuracy digital caliper, weighed, photographed, and reproductive status was determined based on the presence / absence of nipples, the size and position of the testicles and the degree of ossification of the metacarpals (^{Morris 1972}). We collected two specimens which were deposited in the mammal collection of Museo de Historia Natural de la Universidad Distrital Francisco José de Caldas, under catalogue numbers MUD 1140 and MUD 1141. The taxonomic identification follows ^{Gardner (2007)} and ^{Díaz et al. (2016)}.

Fig. 1 Location of the roosts of Lophostoma brasiliense in the Santa Librada Agroecological Reserve, El Líbano, department of Tolima, Central Andes in Colombia. 

Diet

The diet was inferred from fecal samples, left by the bats in the cloth bags and collected under each shelter. Parts of insects collected were separated, identified and quantified by the number of wings, legs and exoskeleton of prey that were found, with the help of a Leica EZ4 HD stereoscope. Parts of arthropod wings were cleaned using a 10%phenol solution, being later dried and photographed. We followed ^{Warren et al. (2013)} for the taxonomic determination and had the collaboration of experts for some confirmations. The food items were identified to the lowest possible taxonomic level, mostly to the level of order. Additionally, we built a reference collection of specimens, mainly of Coleoptera, Diptera and Lepidoptera, that were captured 10 meters around the shelter’s sites, for later comparison with the fragments found in the feces. For this proposal, we used Van Someren-Rydon, entomological nets and light traps. We also include data about composition and abundance of insects present in SLAR, that were collected between 2017 and 2018 (^{Valenzuela et al. 2019}). The frequency of consumption of each prey item was calculated by the number of parts containing such item over the total number of analyzed samples.

Activity paterns

Once identified, a female and a male that were located underneath a house’s basement, were studied by direct observations, in order to know the activity patterns of these individuals. The observations were made during three nights (January 15-17, 2018, during the Moon’s new phase), in which two researchers seated 10 meters away of the bat’s shelter-that had a unique exit-at different places, they counted the number of entry and exit of the bats to the shelter each night, as well as the duration of these movements. The activities were classified, according to the number of departures, as follows: low activity (less than five exits per hour), average activity (between five and 10 exits per hour) and high activity (more than 10 exits per hour). It was not possible to discriminate the number of exits for each sex (male and female).

Bats and Termite nests database compilation

To determine the frequency of use of this type of shelter, among Chiroptera, we conducted a literature review on termite nests used by bats, by using several tools, such as Google Scholar (https://scholar.google.com/), ISI Web of Knowledge (http://www.webofknowledge.com), Scopus (https://www.scopus.com), Biodiversity Heritage Library (https://www.biodiversitylibrary.org/) and institutional repositories. For these searches, we used keywords in Spanish, English and Portuguese for the terms "termites", "termite mounds", "ant nests", "termite nests", "termitaria", "use of termite nest", "bats", "Chiroptera", and "Lophostoma". Our compilation also included unpublished roosting data and photographs of different authors which unambiguously allowed the identification of these species.

RESULTS

Roosting ecology of Lophostoma brasiliense

We collected information on the roosting ecology from three L. brasiliense colonies, two of them occupying termite nests and one roosting on a shelter (underneath house’s basement). All shelters were found inside human-made structures.

Shelter 1: On January 25^th, 2016, we recorded seven individuals of L. brasiliense hanging inside an active termite nest. The termite nest built by Nasutitermes spp. (Isoptera: Termitidae), was 57 cm height, 44 cm width, presenting two cavities. The first cavity, which was used by bats, was 28.8 cm deep with 9 cm of diameter, while second cavity, not occupied by bats, was 24 cm deep with a diameter of 5.8 cm (Fig. 2). This termite nest, at 2.5 meters above ground, was located on the roof of an abandoned house. Bats were not the only animals using the termite nest: juveniles and adults of Blaberus giganteus (Blattodea: Blaberidae) cockroach occupied the second cavity, while the Pheidole spp. (Hymenoptera: Formicidae) was occupying both cavities of the nest. The presence of Pheidole spp. and Nasusitermes spp., seemed not to bother the bats, which even had some of these organisms in their bodies. The social group of L. brasiliense consisted of four females (three adults and one juvenile), two males (adult and juvenile) and another individual that we were not able to capture. Two years later, we visit again this termite nest, but termitaria did not have bats, termites or ants.

Shelter 2: We found this shelter on January 12^th, 2018. It was occupied by three individuals and located in an abandoned house, at two meters above ground. This termite nest showed no signs of the presence of termites or ants. It was 50 cm high and 42 cm wide, and had a single cavity with 5.1 cm diameter and 20 cm depth. Bats shared this cavity with a juvenile Blaberus giganteus. This social group of L. brasiliense was composed of two adult females, and one adult male.

Shelter 3: We captured and identified two adult bats, a female and a male, of L. brasiliense in a shelter that was located underneath house’s basement, on January 15^th, 2018. Bats were hanging at <40 cm distance from the ground. This site only had one exit and it was the only shelter located in this inhabited house.

In Table 1, we summarize these termite nests measurements and compared them to the measurements made by ^{Kalko et al. (2006)} and ^{York et al. (2008)}.

Diet of Lophostoma brasiliense

From analyses of fecal samples, we found remains of wings, legs, antennae, mouth parts, and other fragments, totaling 185 parts of preys, corresponding to six orders of insects. Coleoptera, Heterocera and Hymenoptera were the most frequently consumed preys (respectively, 26.5%, 20% and 14.6%), while Blattodea and Orthoptera were the least consumed (Table 2). Dietary composition included those insects species more close to the nest with a high frequency of beetles (Melolonthidae) and moths (Heterocera).

Activity paterns

We recorded a moderate activity between 17:30 and 19:30 h. Bats’ activities were low, with five exits in this period, and the individual flights out of the termite nets did not last more than five minutes. Between 19:30 and 00:00 h the activity dropped even more and we only observed two sporadic flight outs. The most intense activity was observed between 00:00 and 02:00 h, with more than 25 exits, each lasting for approximately two minutes. After this time, activity was low, with less of five flight outs until dawn.

Bats and use of termite nest

We compiled a total of 55 records of termite nests used by 15 bat species from five families in 25 localities around the world, including our own data. Our compilation showed that termite nests were used by bats from the families: Phyllostomidae (10 species), followed by Emballonuridae (2), Vespertilionidae (1), Pteropodidae (1), and Nycteridae (1). It is noteworthy that 87.3% of the records is associated with species of subfamily Phyllostominae, and 78.2% is related to species of the genus Lophostoma (L. brasiliense, N = 19; L. carrikeri, N = 2; L. evotis, N = 2; L. kalkoae, N = 1; L. silvicolum, N = 19), showing a high preference of such shelters by species of this genus (Table 3).

DISCUSSION

The nests we observed have similar measurements to those reported by other authors. Our termites nest were located at a similar height, had the same depth, but were smaller and narrower than those reported by ^{Kalko et al. (1999)} in Peru. ^{York et al. (2008)} did not offer measures of the termitaria in Costa Rica and only described a termite nest with about 20 cm of depth and therefore shallower than ours.

Fig. 2 Termite nest used by Lophostoma brasiliense in Central Andes, El Líbano-Tolima, Colombia. A. Bats roosting inside termite nest, B. Termite nest in abandoned houses roof and C. Graphic representation of termite nest with bats inside. 

Table 1 Termite nest measurements used by Lophostoma brasiliense. 

L. brasiliense has been reported using termite nests of Microcerotermes arboreus and Nasutitermes spp. (Isoptera: Termitidae) in Trinidad and Costa Rica respectively (^{Goodwin & Greenhall 1961}) and in shelters located between 1.5-2.5 meters above ground as recorded by several authors (^{Handley 1966}; ^{Peracchi & Albuquerque 1993}; ^{York et al. 2008}).

Records of individuals of L. brasiliense roosting in termite nests vary from a single individual to groups with up to five individuals. ^{Peracchi & Albuquerque (1993)} and ^{York et al. (2008)} found five individuals occupying the same nest at once, while ^{Rojas-Rojas-Rojas et al. (2015)} and ^{Goodwin & Greenhall (1961)} found groups of four individuals. Here, we reported the largest group currently found with seven individuals occupying a same nest which is remarkable due to space constraints offered by termite nest.

Activity patterns of L. brasiliense are little known (^{Mangolin et al. 2007}) and have been hypothesized to be similar to L. silvicolum patterns, which has a somewhat sedentary behavior and appears to stay close to its roost and foraging at distances from 200 to 500 m (^{Kalko et al. 1999}). Our observations suggest that L. brasiliense may also be a sedentary hunter, moving away from its roosts for short distances. ^{Mangolin et al. (2007)} reported a high and constant activity of L. brasiliense late at night, when the bats used to leave their roosts for brief periods, whose were shorter than those reported to the species L. silvicolum by ^{Kalko et al. (1999)}.

Termitaria as nest sites for bats

Our compilation on the use of termite nests by bats is the first review on the subject. We have shown that most bats using termite nests are insectivores (74%), which contrasts with the prevalence of the use of modified and unmodified leaves for roosting in frugivorous bats (^{Garbino & Tavares 2018}). Some bats can use termite nests opportunistically since this behavior has been reported for some species, such as the fruit-eating bat Artibeus fraterculus (^{Carrera et al. 2010}; ^{Hernández-Mijangos 2010}), the carnivore bat Chrotopterus auritus (^{Sanborn 1932}; ^{Medellín 1989}), the insectivore Micronycteris megalotis (^{Patterson 1992}), Phyllostomus hastatus (^{Voss et al. 2016}), the emballonurid species Saccopteryx canescens, and S. leptura (^{Ibañez 1981}), and the old world species Murina ßorium (^{Clague et al. 1999}) and Balionycteris maculata (^{Hodgkison et al. 2003}).

Table 2 Diet of Lophostoma brasiliense in Santa Librada Agroecological Reserve, El Líbano-Tolima, Colombia. 

Nesting inside termitaria can offer advantages for bats, such as protection against predators and a suitable micro-climate (^{Boonman 2000}). However, not every termite nest is occupied, although this resource can be abundant or scarce. ^{Villalobos-Chaves et al. (2016)} observed a low rate of occupation of termites nests in Costa Rica (1 out of 74 nests) and ^{Mangolin et al. (2007)} only recorded two occupied termite nests out of ten in Brazil. In our study, after two visits (2016 and 2018) each one with 30 sampling days (unpublished data), we did not find more termite nests, despite the intensive searches that we carried out, suggesting the possibility that hese types of shelters are scarce in our study area. Also, the low occupation rates of termite nests can be due to low population densities and high selectivity of termite nests for this species. ^{Kalko et al. (2006)} showed that the quality of roosts, including the physical structure of the termite nests, is key to roosting choice by the bats.

Although there are some reports on the occupation of tree holes (^{Handley 1966}), small cavities in rocks (^{Robinson & Lyon 1901}) and human-made structures -as reported by ^{Hice et al. (2004)} and this work- the majority of data available (see Table 3) point to a noteworthy preference of termite nests by L. brasiliense.

Therefore, although at a first glance, the occupa tion of termite nests appears to be opportunistic, we herein compiled evidence to reinforce the occurrence of unique relationships between species of Lophostoma and termite nests. This rare roosting behavior of Lophostoma species in termite nests could represent constraints for species conservation, related to the very specific requirements regarding food or shelter (^{Sagot & Chaverri 2015}) that may be enhanced by threats and concurrent effects, as given by habitat loss and fragmentation.

Table 3 Compilation on reports of bats using termite nests as roosts worldwide. 

CONCLUSIONS

We reviewed evidence of the strong association between termite nest and Lophostoma species and we provide new data on the roosting ecology of L. brasiliense. Future studies should consider other interactions between bats and termites, not addressed in this review, as for example chemical responses of termites against the modification and use of its ter mitaria and the possible influence of termite species in bat distribution. Further studies could help to elucidate the evolutionary history and distribution patterns of the genus.