Acoustic data collection was performed in March 2009, in La Gamba, southwestern Costa Rica. Bats were captured by searching suitable leaves; upon capture, bats were housed in cloth bags and fitted with numbered metal wing bands. A suitable tubular leaf was then located near the capture site to perform acoustic trials. If suitable leaves were not found, we cut an unoccupied tubular leaf from another site and planted it near the capture site. We conducted all trials in areas where bats would have enough space for manoeuvrability and access to the experimental leaf.
Three condenser microphones (CM16, Avisoft, Berlin, Germany) mounted on tripods were used to record calls produced in the release area; two microphones were located less than 5 cm from the tubular leaf at different heights, close to where bats would be roosting, and the other microphone was placed on the path between the release site and the tubular leaf. Release sites were usually located at approximately 5 m from the experimental leaf. Social calls were recorded onto a Dell Latitude laptop through Avisoft's UltraSoundGate 416 and Avisoft-Recorder software (sampling rate 384 kHz, 16-bit resolution).
Trials began by placing one individual from the captured group inside the experimental leaf (i.e. roosting bat), and then placing a second bat from the same group in the researcher's hand at the release site (i.e. flying bat). The cloth bag with the remaining bats was placed inside a waterproof bag to avoid acoustic interference. Recordings began when we released the flying bat, and ended when the bat entered the experimental leaf or after the flying individual left the area and was not visible for more than 1 min.
For every acoustic trial we noted the identity of the roosting and flying individuals, and whether the flying individual entered the leaf. Upon later analysis, we also noted if calls were emitted by the flying individuals (‘inquiry calls’), and/or the roosting individuals (‘response calls’). We ran two G-tests of independence to determine (i) if the presence of a response call was related to the presence of an inquiry call; and (ii) if entrance of the flying individual into the roost was related to the production of response calls by the roosting individual. Call characteristics were measured using Avisoft SasLab Pro. For both call types, we measured call duration (Dur), frequency at call start (Fstart), frequency at call end (Fend), peak frequency (Fpeak) and bandwidth (BW). We also gauged how rapidly bats responded by measuring the time from the start of an inquiry call to the start of the corresponding response call. Since multiple response calls were emitted in quick succession (i.e. a bout), we also determined the average number of calls per bout and bout duration.