General description of the method
To investigate the foraging behavior of mother–pup pairs, we arranged artificial flowers in the environment of the bats’ day roost and tested for communal and independent visits of mothers and their pups. Each flower comprised a RFID reading system that allowed us to precisely record number and time point of visits of mothers and their pups, which were both tagged with RFID transponders.
Capturing, captivity period, and tagging of mother–pup pairs
During two reproductive periods in 2016 and 2017, we caught 18 lactating female G. soricina together with their non-volant pups in the national park Santa Rosa, Costa Rica (UTM: 16P 651137 1198498). Mother–pup pairs were caught with a hand net inside their day roost (16 pairs) or with mist nets (Monofilament, Ecotone, Gdynia, Poland) in front of their presumed roost (2 pairs). Bats were identified as G. soricina following the field key by Timm and LaVal (1998). Capturing bats while pups were still non-volant and attached to their mother’s teat during capture was a crucial precondition to allow a clear assignment of pairs. Prior to the field experiment, mother–pup pairs were kept in flight cages for a captivity period of 34.7 ± 10.2 days (11.7 m2; Hexagon Screen House; Eureka) and fed with a diet based on NektarPlus (mixing ratio 1:5 in tap water; Nekton GmbH, Pforzheim, Germany) offered in cylindrical bird water feeders with a protruding opening. In 2017, we additionally offered a honey in water solution that we partially enriched with bee pollen and milk powder (NAN Confort Digestivo 2, Nestlé).
After pups became volant and later proficient in feeding on the feeders within the flight cage, we removed feeders and familiarized all bats with one of the artificial RFID flowers (Fig. 1a and detailed information below) that we later used in the field experiment. Mothers and pups fed from this new flower type inside the flight cage for at least two nights before they were released to the field experiment.
Bats were tagged with 125 kHz RFID glass tube transponders (E675-313-Uni, I-KEYS RFID-Technik, Berlin, Germany) that we glued to the slightly sheared upper back between the scapulae using a flexible staying skin bond (SAUER-Hautkleber type 50.22, Manfred Sauer GmbH, Lobbach, Germany). Transponders were covered with a black shrinking tube to protect bats in the highly unlikely case of a destroyed glass tube. Including this cover, transponder weight was 0.28 g and was therefore even for pups less than 5% of their body weight, which is suggested not to alter maneuverability (Aldridge and Brigham 1988). Bats were observed to perform increased grooming activities only for a short time after being tagged, but transponders seemed not to have any lasting alteration effects on bats’ behavior and adhered for at least 1 week before they fell off.
To ensure that all bats were proficient in feeding on our artificial flower inside the flight cage, we logged bat visits for at least one night before the field experiment started with releasing the tagged bats. Mother–pup pairs that had been captured inside their day roost were released at the same location during the afternoon, and mist-netted pairs were released at the place of capture after sunset.
In the field experiment, we tested for communal and independent visits at artificial flowers that we arranged around the bats’ day roosts. We used custom-made artificial flowers that comprised a photoelectric through-beam sensor and a RFID reading system, which allowed us to precisely record time and number of visits, as well as the identity of each RFID-tagged individual (Fig. 1a). Computational hard- and software of flowers was based on the Arduino platform and consisted of a microcontroller board (2016: Arduino UNO clone; 2017: Arduino Nano clone; ATmega328P), a RFID unit (RDM6300 v. 2.0), a Real-time-clock (2016: DS1307; 2017: DS3231), and a SD-Card reader. Each flower comprised a nectar reservoir, which was filled with sugar water of 17% sucrose (azúcar refinado, Victoria, LAICA, Costa Rica). Flowers held enough nectar to never get depleted during an experimental night. To reach the nectar, bats had to insert their head into the protruding rectangular flower opening (3 × 4 cm). As all outer parts of the flower, this opening was made out of plastic and carried the RFID antenna and the photoelectric through-beam sensor. The flower opening was slightly inclined downward, roughly resembling the bell-shaped flower type of the chiropterophilous Crescentia alata (Porsch 1931) that is common in the study area. We mounted flowers on string lines in heights of 1.2 m to 1.9 m above ground, which were in similar heights as many flowers of two abundant chiropterophilous plants (Crescentia alata and Bauhinia ungulata). However, there are also other plant species occurring in the area that produce flowers much higher above ground (e.g., Pachira quinata).
In 2016 (setup “Casona 2016”), our RFID flowers were arranged at a G. soricina day roost located inside a room of the historical hacienda “La Casona” in Santa Rosa National Park. This roost was used by a variable number of ca. 20 to 25 individuals. We caught seven mother–pup pairs, of which five pairs were caught with a hand net inside, while two were mist netted in front of the roost. While keeping these bats in a flight cage, mother–pup pairs were additionally used for an experiment on maternal mouth-to-mouth feeding behavior (Rose et al. 2019). Following this captivity period, all seven mother–pup pairs were released to the field experiment on 25.02.2016. However, from these pairs, we had to exclude two from the further analysis, as their pups were found dead two days after release, one trapped in a cleaning bucket in front of the day roost and the second one on the floor inside. We placed four RFID flowers in the close environment outside the roost (5–60 m distance): Two were located close to the exterior wall of the historical hacienda, and the other two RFID flowers were mounted in the surrounding forest (Fig. 1b). RFID flowers were active during three consecutive nights, but one flower was not logging visits during the first night due to a technical problem.
In 2017 (setup “Casona 2017”), we caught ten mother–pup pairs from the same day roost as in 2016. All pairs were caught with hand nets inside the roost. Following the captivity period, seven pairs were released on 05.02.2017 and three pairs on 11.02.2017. RFID flowers were arranged at the same locations as in 2016, but we added one additional flower in the forest near the roost, and four additional RFID flowers at places further away (380–874 m distance): two were mounted at bat-pollinated Crescentia alata trees, one close to a flowering bat-pollinated Bauhinia ungulata, and one was placed at the location of the flight cage where bats were temporarily kept during the captivity period (Fig. 1c). Flowers were active from 05.02.2017 until 17.02.2017. Two flowers were first mounted on 06.02.2017. On 12.02.2017, the nectar reservoir of one flower was hanging down in the morning and was probably not available to the bats for most of this night.
One additional mother–pup pair was caught in 2017 within another day roost, a storeroom of the research station in the park (setup “Bodega 2017”). The pair was released on 23.02.2017. In this setup, we arranged four RFID flowers close to the roost (9–65 m) and four further away (105–244 m) (Fig. 1d). Three of the RFID flowers were mounted at the same positions as previously used in the “Casona 2017” setup. Flowers were active from 22.02.2017 until 13.03.2017.
In total all three experimental setups in 2016 and 2017 provided 34 observation nights with a mean number of 7.9 ± 1.4 active RFID flowers, summing up to a total number of 3197.8 RFID flower hours. Overall, we analyzed the behavior of 16 mother–pup pairs with 4.0 ± 3.6 pairs per observation night.
Other RFID-tagged bats
To gain additional data on the foraging behavior and RFID flower usage of unrelated G. soricina and for checking proper function of our setup, we additionally RFID tagged 25 other individuals. In the experimental setup “Casona 2016”, we additionally tagged three adult males that were released together with the mother–pup pairs and that were equally familiarized with the RFID flowers before. In the “Bodega 2017” setup, we additionally tagged 22 individuals (10 adult males, 7 adult females, and 5 juveniles/subadults). One adult male was caught together with the mother–pup pair in their day roost and treated equally. The other individuals inhabited unknown day roosts and were mist netted in different nights at varying locations in vicinity to RFID flower positions. Six individuals were familiarized over two nights with the RFID flowers in flight cages before release, while 15 individuals were released immediately after being equipped with transponders.
Statistical analysis of independent or communal foraging of mothers and pups
For all tagged bats, we analyzed the time, location, and number of visits at our RFID flowers. We excluded RFID reads without associated photoelectric sensor events, as they represented unsuccessful approaches without entering the flower opening, or derived from bats using flowers as temporary perches. A new visit was counted after a pause of at least one second without any sensor event. Visits of mothers and their pups were classified as communal if visits were performed at the same RFID flower position with less than 60 s interval (either mother or pup visiting the flower first). Otherwise, visits were classified as independent. In case pups discovered a RFID flower position independently from their mother, we additionally analyzed all first encounters to check whether they may have followed unrelated conspecifics.
Photoelectric sensor events without RFID reads were counted as visits of untagged bats. However, this category was of limited reliability since sensors were occasionally also triggered by insects like ants or might have partially derived from tagged bats in case of a failed RFID transmission to the flower. As a consequence, we excluded events after sunrise as well as events of less than 0.1 s duration, as they were most likely not representing feeding bats.
Statistical analysis was performed in R (v. 3.4.3, R Core Team 2017) using the Rcmdr package by Fox and Bouchet-Valat (2017). Differences in the number of RFID flower visits between mothers and pups were tested using non-parametric Wilcoxon rank-sum test. Numbers of independent and communal visits of mother–pup pairs were compared pairwise by Wilcoxon signed rank test. All tests were performed two sided. Data from both years were pooled for analyses. If multiple tests were performed on the same dataset, we adjusted significance level (α = 0.05) using sequential Bonferroni correction. Maps for spatial visualization of RFID flower positions were created using the get_map function in the ggmap package by Kahle and Wickham (2013).
Observations on pup development during the captivity period
Pup growth was documented by taking measurements of each individual at least at the day of capture and before releasing them to the field experiment. We measured length of forearm (FA) with a caliper to the nearest 0.1 mm and body mass (BM) with a spring balance to the nearest 0.5 g. Identification of individuals and assignment of pairs were realized by marking mothers with color-coded collars and by shearing small parts of the fur at identical locations in mothers and pups or by perforating wing membranes at same locations with a small biopsy punch (2 mm). In 2017, nursing behavior of mother–pup pairs during daytime was documented at least on the day before bats were released to the field experiment with a camcorder or by taking photographs.