Dynamic Echo Information Guides Flight in the Big Brown Bat

Abstract

Animals rely on sensory feedback from their environment to guide locomotion. For instance, visually guided animals use patterns of optic flow to control their velocity and to estimate their distance to objects (e.g., Srinivasan et al., 1991, 1996). In this study, we investigated how acoustic information guides locomotion of animals that use hearing as a primary sensory modality to orient and navigate in the dark, where visual information is unavailable. We studied flight and echolocation behaviors of big brown bats as they flew under infrared illumination through a corridor with walls constructed from a series of individual vertical wooden poles. The spacing between poles on opposite walls of the corridor was experimentally manipulated to create dense/sparse and balanced/imbalanced spatial structure. The bats' flight trajectories and echolocation signals were recorded with high-speed infrared motion-capture cameras and ultrasound microphones, respectively. As bats flew through the corridor, successive biosonar emissions returned cascades of echoes from the walls of the corridor. The bats flew through the center of the corridor when the pole spacing on opposite walls was balanced and closer to the side with wider pole spacing when opposite walls had an imbalanced density. Moreover, bats produced shorter duration echolocation calls when they flew through corridors with smaller spacing between poles, suggesting that clutter density influences features of the bat's sonar signals. Flight speed and echolocation call rate did not, however, vary with dense and sparse spacing between the poles forming the corridor walls. Overall, these data demonstrate that bats adapt their flight and echolocation behavior dynamically when flying through acoustically complex environments.

Keywords: animal flight; corridor; echo flow; echolocation; sonar.

FIGURE 1

Cartoon illustrating the concept of echo flow as used in the present article. (A) Bat (black, t₁) enters the tunnel and emits a vocalization (blue). Echoes return to the bat from the left (orange) and right (green) corridor sides. In the time that the bat emits the call (blue) it moves (gray), thereby slightly displacing the vocalization (light blue) and creating complex echo patterns. By the time the bat sends out its second call (black dashed line, t₂), it has traveled further down the corridor, but some echoes from the previous call (t₁) may still be returning to the bat. These echoes will overlap with the subsequent call emission (blue, t₂) and echoes returning from call 2. (B) Biosonar vocalizations illustrated as sounds waveforms (blue) and echoes returning to bat from poles located at different distances (orange) from both corridor walls (orange vs. green) are shown across time (x-axis). Gray echo illustrates the echo from the wall at the end of the tunnel (see Figure 2A). The onset of call 2 (call at t₂) occurs when echoes from call 1 (call at t₁) are still arriving (dashed black arrow). The distance that the bat has flown in between two calls in schematically plotted on the (y-axis). Calls (blue) and echoes (orange/green/gray) are matched between panels (A) and (B). Note that head and ear movements are not displayed and may further complicate echo patterns. (C) Schematic of waveforms across time (x-axis) as the bat flies down the corridor and emits successive vocalizations (blue). Pulse-echo overlap may occur (e.g., call 2, 4, and 6), complicating the spectral and temporal echo structures. The wall echo (gray) will increase in intensity and move closer to the time of vocalization as distance to the end of the corridor decreases. Y-axis plot distance flown between successive calls. Note that waveforms in (B) and (C) are cartoons illustrating the complex merging of the echoes returning to the bat; they do not represent how the bat perceives the flow of echoes.

FIGURE 2

Experimental procedures. (A) Cartoon of the 6 m × 6 m × 2.5 m flight room, covered in acoustic foam, housing a corridor (620 cm × 120 cm) created from individual wooden poles (gray circles) so that the corridor’s left (L) and right (R) walls could be independently manipulated. Bats had been trained to fly through an elliptic hole (inset labeled “frontal view”) embedded in a rectangle of acoustic foam and black felt to prevent them from gaining echo information about the pole spacing prior to entering the corridor. The felt extended to the walls of the flight room (black line labeled felt curtain). Thirteen high-speed IR motion-tracking cameras were mounted within the tunnel (omitted for visual clarity) to capture the bat’s flight path, pole and microphone positions. Four ultrasonic microphones (m) were mounted vertically along the height of the room at the end of the tunnel. (B) Each of the seven bats flew at least ten trials in each of the following four different tunnel configurations. For each configuration “dense” refers to a 12 cm spacing of wooden poles, “sparse” refers to a 36 cm spacing of poles. (1) Left dense, right dense (LD-RD), (2) Left sparse, right sparse (LS-RS), (3) Left dense, right sparse (LD-RS), (4) Left sparse, right dense (LS-RD). (C) Photograph of the actual head marker, measuring 3.5 cm in length and 0.5 cm in width (original background grayed out) on the bat. Each reflective ball (m 1, m 2) to be tracked by a motion-tracking system is 8 mm in diameter. Head marker weighed 0.9 g total, making up about 5–7% of the animal’s body weight. Marker was attached using water-soluble theater glue. (D) Spectrogram illustrating sonar sound groups and call types as defined in the present study. Time series of calls is plotted as frequency (kHz, y-axis) across time (seconds, x-axis). For each call, the call type is indicated in the upper white box, while each sonar sound group type is indicated in the lower box. For sound groups, box shape and color correspond to marker shape and color in Supplementary Figure S2. Codes: s, single call/PI; d, doublet call/PI; t, triplet call/PI; q, quadruplet call/PI; p, pre-/post-sound group PI.

FIGURE 3

Distribution of flight path deviation from the midline across conditions. Histogram of the distribution of deviation (x-axis) from the midline (black dashed line) across all conditions (LD-RD, LD-RS, LS-RD, and LS-RS). Data are plotted in 2.5 cm bins, and balanced pole-spacing conditions (LD-RD and LS-RS) are illustrated in light gray, imbalanced pole–spacing conditions (LD-RS and LS-RD) are illustrated in dark gray. Average deviation from midline is indicated by a black triangle. Bats’ flight path differs significantly between acoustically unbalanced conditions (LS-DR and LD-SR), and unbalanced conditions differ significantly from balanced conditions. Balanced conditions (LD-RD and LS-RS) do not differ significantly.

FIGURE 4

Distribution of raw flight tracks across conditions and bats. For each condition (right y-axis) raw flight tracks (gray) and their mean (red) are plotted as distance from the corridor end (lower x-axis). Tracks are plotted as deviation from the midline (black dotted line) along the left y-axis. The letter in the corner of each condition indicates the spacing of poles on that side of the condition. All data are plotted for each of the seven bats (upper x-axis). Bats’ flight paths steer away from densely spaced corridor walls and center otherwise.

FIGURE 5

Flight speed, call rate and call duration across conditions. (A) Mean flight speed ±1 SE in m/s (left y-axis), is plotted across conditions (x-axis). When the animals fly though the different corridor manipulations (LD-RD, LS-RS, and S/D) flight speed (circles, solid line) is relatively stable at around 3.8 m/s with no significant differences across conditions. Also plotted is the mean call rate per second ±1 SE (right y-axis) across conditions (squares, dashed line). There are no significant differences in call rate in baseline conditions, though the difference between S/D and both baseline conditions is significant. (B) Mean call duration ± SE (y-axis) is plotted for each condition (x-axis). Bats use significantly shorter calls in the LD-RD condition compared to all other conditions (LS-RS, S/D). Asterisk indicates significance at p = 0.05 level.

FIGURE 6

Sonar sound groups and pulse interval distributions. (A) Proportion of calls that are either single calls (single) or part of a sonar sound group (doublet, triplet, and quadruplet) is plotted on the y-axis across conditions (x-axis). Around 62% of all calls are doublets. (B) The mean pulse interval ±1 SE (PI, y-axis) that bats use when navigating the different corridor configurations show differences across sound group types (x-axis). Pulse intervals between single sounds (see red “s” in Figure 2D) are significantly longer, around 60 ms. By contrast, doublets, triplets and quadruplets share a shorter PI around 40 ms. Data are pooled across conditions, as no difference across conditions (LD-RD, LS-RS, and S/D) was found. Asterisk indicates significance at p = 0.05 level.