Multimodal cue integration in the dung beetle compass

Abstract

South African ball-rolling dung beetles exhibit a unique orientation behavior to avoid competition for food: after forming a piece of dung into a ball, they efficiently escape with it from the dung pile along a straight-line path. To keep track of their heading, these animals use celestial cues, such as the sun, as an orientation reference. Here we show that wind can also be used as a guiding cue for the ball-rolling beetles. We demonstrate that this mechanosensory compass cue is only used when skylight cues are difficult to read, i.e., when the sun is close to the zenith. This raises the question of how the beetles combine multimodal orientation input to obtain a robust heading estimate. To study this, we performed behavioral experiments in a tightly controlled indoor arena. This revealed that the beetles register directional information provided by the sun and the wind and can use them in a weighted manner. Moreover, the directional information can be transferred between these 2 sensory modalities, suggesting that they are combined in the spatial memory network in the beetle's brain. This flexible use of compass cue preferences relative to the prevailing visual and mechanosensory scenery provides a simple, yet effective, mechanism for enabling precise compass orientation at any time of the day.

Keywords: insect; navigation; sun; vision; wind.

Fig. 1.

Wind conditions in the natural habitat of the ball-rolling dung beetle S. lamarcki. (A) Left axis (blue): Average local wind speeds over the course of a day (n = 5). Shaded blue area shows the minimum and maximum measured wind speeds. Right axis (red): A normalized activity diagram of S. lamarcki (data adapted from ref. 26) over a 24-h period. (B) The normalized frequency of angular change in wind direction over a 6-min period. As S. lamarcki roll their balls for only ∼6 min on average, we calculated how much the wind direction changed within this time window. Wind direction data were extracted from 06:30–18:30, when the diurnal dung beetle S. lamarcki is most active (see A). The median change in wind direction within a 6 min window is 27.0° ± 48.5° (median ± IQR).

Fig. 2.

The cue preference of the wind and sun compass at different solar elevations. (A) Schematic illustration of the experimental procedure. Two wind stimuli (open circles with blue, curved arrows) were set 180° apart and perpendicular to the sun position (yellow circles). During their first roll, the beetles experienced the sun (and other celestial cues) and wind from a certain direction. In the control conditions, both cues remained in those positions during a second roll. In the test conditions, one cue, either the wind or the sun was displaced by 180°. The change in heading between the 2 consecutive rolls was calculated. (B–D) Circular plots showing the change in heading between 2 consecutive rolls. Each dot indicates the change in heading between 2 rolls performed by the same individual. The black (control), blue (test, 180° change in wind direction), or yellow (test, 180° apparent change in solar azimuth) lines represent the mean change in heading (μ). The associated sector represents the 95% confidence interval (CI) of the data. Upper plots: the change in heading of beetles (n = 30, each) when the wind was switched at high [≥75°] (B), medium [45–60°] (C), and low [15–30°] (D) sun elevations (test, blue circles) or when the wind was held in place between the 2 rolls (control, open circles). Lower plots: Change in heading of the same beetles when the sun was switched ∼180° (test, yellow circles) at high (B), medium (C), and low (D) elevations (controls without manipulations are shown as open circles). (E) Initial rolling directions of the tested beetles with respect to the position of the main cue used at a given elevation. Heading direction relative to the wind at high sun elevation (Upper histogram), and with respect to the wind stimulus at medium sun elevation (Middle histogram), and low sun elevation (Lower histogram). (F) The beetle’s ability to maintain a straight course (n = 25) was tested in the absence (sun, yellow box plot) and presence (sun and wind, blue box plot) of wind at high sun elevation (each beetle exited the arena 10 times with, and 10 times without a wind stimulus). The mean vector lengths (precision) for each beetle were significantly greater when wind was present compared with the orientation performance without wind. Black squares indicate the average and black horizontal lines the median lengths. The boxes show the interquartile range and whiskers extend to the fifth and 95th percentile. *P < 0.05.

Fig. 3.

Testing the multimodal beetle compass indoors. (A) Schematic of the indoor arena. An ersatz sun comprising a single green LED was angled at an elevation of 45° to the arena’s center. Artificial wind at 90° to the ersatz sun was produced using 2 fans placed opposite one another. The beetle’s behavior was recorded by a videocamera positioned directly above the center of the arena. (B) We tested if beetles without (and with) antennae are able to follow a 180° switch of the wind stimulus origin. A ventral view of a beetle’s head capsule is shown to indicate the location of the antennae. Each beetle experienced wind during the first roll and a wind stimulus at an angular difference of 180° during the second roll. In beetles with antennae, the change in heading was significantly clustered around 180° (Left plot, blue circles), which was not the case in beetles without antennae (Left plot, black circles). Not surprisingly, the absolute change in heading was larger in the beetles with antennae intact (Right plot, blue box plot) than in the beetles without antennae (Right plot, black box plot). In the Left plot, the line shows the mean direction and the sector indicates the CI. In the Right plot, the white squares indicate the average, and white horizontal lines the median absolute change. The boxes show the interquartile range and whiskers extend to the 5th and 95th percentiles. *P < 0.05. (C) Change in heading between 2 consecutive rolls of individual beetles when presented with an ersatz sun and artificial wind during both rolls (Left plot, black circles) or when one cue, either the ersatz sun (Middle plot, green circles) or the wind (Right plot, blue circles) were the only orientation reference during the second roll (while both cues were available during the first roll). In all cases, the change in heading is significantly clustered around 0°. The mean directions (μ) are indicated by black, green, or blue lines, respectively. The sectors indicate the CIs.

Fig. 4.

Transfer of directional information between the sun and wind compass. (A) To investigate whether the sun and wind compass are integrated into the same neuronal circuitry in the central brain, we performed indoor transfer experiments. The beetles set their direction with respect to the ersatz sun during their first roll. After another 2 rolls, in which the beetles experienced both modalities (ersatz sun and wind), the beetles experienced wind as their only source of reference in their fourth roll (or vice versa, from wind to sun). The change in heading between the direction taken during the first and fourth rolls was significantly clustered around 0° (V test). (B) We tested the same transfer experiment within the visual modality, by positioning a UV light spot 180° to the ersatz sun. Similar to the experiments in A, the change in heading between the first and fourth rolls was significantly clustered around 0° (V test). The lines in A and B show the mean direction, the sectors indicate the CIs. (C) To further evaluate the statistical results of the experiments in A and B, we randomly shuffled the heading directions obtained in our experiments 100,000 times and simulated a distribution of mean absolute change in headings. In both cases, for the data with 2 modalities (Left plot) and the experiments within one modality (Right plot), we found that the mean absolute changes in heading obtained (red lines) differed significantly from the simulated mean absolute change in heading distribution (gray histograms).