Honeybee Navigation En Route to the Goal: Visual Flight Control and Odometry

The precision with which a fly lands on the rim of a teacup, the reliability with which a honeybee repeatedly returns to a nectar-bearing patch of flowers and the ease with which a foraging Saharan desert ant finds its way back home indicate that insects, despite their relative simplicity, possess visual systems that are capable of high performance. How is this performance achieved? Some of these strategies are described below, in relation to specific tasks of navigation and flight control that would appear to tax a small nervous system which carries fewer than one-hundredth of 1 % as many neurones as the human brain.

When a bee flies through a hole in a window, it tends to fly through its centre, balancing the distances to the left and right sides of the opening. Bees, like most insects, possess very small interocular separations and therefore cannot rely on stereoscopic vision to measure distances to objects or surfaces (Collett and Harkness, 1982; Horridge, 1987; Srinivasan, 1993). How, then, does the bee fly through the middle of the hole? One possibility is that it simply balances the speeds of image motion on the two eyes as it flies through the opening. To investigate this hypothesis, we trained bees to enter an apparatus which offered sugar solution at the end of a tunnel 40 cm long, 20 cm high and 12 cm wide (Kirchner and Srinivasan, 1989). Each side wall carried a pattern consisting of a vertical black-and-white grating of period 5 cm (Fig. 1A). The grating on one wall could be moved horizontally at any desired speed, either towards the reward or away from it. After the bees had received several rewards with the gratings stationary, they were filmed from above as they flew along the tunnel. When both gratings were stationary, the bees tended to fly along the midline of the tunnel, i.e. equidistant from the two walls (Fig. 1A). But when one of the gratings was moved at a constant speed in the direction of the bees’ flight – thereby reducing the speed of retinal image motion on the eye viewing the moving grating relative to the other eye – the bees’ trajectories shifted towards the side of the moving grating (Fig. 1B). When the grating moved in a direction opposite to that of the bees’ flight – thereby increasing the speed of retinal image motion on that eye relative to the other – the bees’ trajectories shifted away from the side of the moving grating (Fig. 1C). These findings demonstrate that when the walls were stationary, the bees maintained equidistance by balancing the apparent angular speeds of the two walls or, equivalently, the speeds of the retinal images in the two eyes. A lower image speed on one eye was evidently taken to mean that the grating on that side was farther away and caused the bee to fly along a trajectory closer to it. A higher image speed, in contrast, had the opposite effect.

Were the bees really measuring and balancing image speeds on the two sides as they flew along the tunnel, or were they simply balancing the contrast frequencies produced by the succession of dark and light bars of the gratings? This question was investigated by analysing the flight trajectories of bees when the two walls carried gratings of different spatial periods (10 cm and 2.5 cm). When the gratings were stationary, the trajectories were always equidistant from the two walls, even when the spatial frequencies of the gratings on the two sides – and therefore the contrast frequencies experienced by the two eyes – differed by a factor of as much as four (Fig. 1D). When one of the gratings was in motion, the trajectories shifted towards or away from the moving grating (as described above) according to whether the grating moved with or against the direction of the bees’ flight (Fig. 1E,F). These results indicate that the bees were indeed balancing the speeds of the retinal images in the two eyes and not the contrast frequencies. The above findings are true irrespective of whether the gratings possess square-wave intensity profiles (with abrupt changes of intensity) or sinusoidal profiles (with gradual intensity changes) and irrespective of whether the contrasts of the gratings on the two sides are equal or considerably different (Srinivasan et al. 1991). In further experiments, the speeds of bees flying through the tunnel were measured and used in conjunction with the known speed of the moving grating to calculate the expected positions of the flight trajectories along the width of the tunnel, on the assumption that the bee balances the angular velocities of the images in the two eyes. The predicted positions agreed well with those observed experimentally (Srinivasan et al. 1991), thus confirming this hypothesis.

Taken together, the above findings suggest that the bee’s visual system is capable of computing the apparent angular speed of a grating independently of its contrast or spatial-frequency content. It is worth noting that, if movement cues are to be exploited to estimate the range of a surface, it is necessary to use a mechanism that measures the speed of the image independently of its geometrical structure. Such a mechanism is what the bee evidently uses to navigate through the middle of a gap between, say, two vertical branches of a tree regardless of the textural properties of the bark on the two sides.

The mechanism that mediates the centring response is quite different from that mediating the well-known ‘optomotor response’. When an insect is placed inside a rotating striped drum, it tends to turn in the same direction as the drum, thereby stabilising its orientation relative to the surroundings. The optomotor response helps the insect to maintain a straight course. The characteristics of the optomotor response have been investigated quite extensively (Reichardt, 1969; Wehner, 1981) and the neural basis of the response is now well on the way to being understood (Hausen and Egelhaaf, 1989). The optomotor response is driven by a mechanism which is sensitive primarily to the contrast frequency of the stimulus and which, therefore, confounds the angular velocity of a striped pattern with its spatial period (Reichardt, 1969). The centring response, however, is mediated by a mechanism that is sensitive primarily to the speed of the stimulus, regardless of the spatial structure of the stimulus or the contrast frequency that it produces. Further experiments, in which bees were exposed to patterns moving in various directions whilst flying through tunnels, indicate that, unlike the mechanism mediating the optomotor response, the mechanism underlying the centring response is insensitive to the direction of image motion (Srinivasan et al. 1993). In this respect, the movement-sensitive mechanism mediating the centring response appears to be similar to that employed by peering locusts to gauge the distance of targets (Wallace, 1959; Sobel, 1990). Additional evidence, obtained using more complex visual stimuli, indicates that the centring response and the optomotor response are mediated by distinct movement-detecting mechanisms (Srinivasan et al. 1993).

In the previous section we showed that image motion experienced by the lateral regions of the two eyes can be used to negotiate narrow gaps or to fly between two potential obstacles. Do bees monitor the apparent velocity of the surrounding environment to control other aspects of locomotion, such as flight speed? The experiments illustrated in Figs 2 and 3 suggest that this is indeed the case. In Fig. 2, bees are made to fly through a tapered tunnel lined with black-and-white vertical stripes to collect a reward of sugar water at the other end. The bees slow down as they approach the narrowest section of the tunnel and accelerate when the tunnel widens beyond it. In fact, the variation of flight speed is very close to that expected if the bees were to hold the angular velocity of the image in the lateral eye region constant as they fly though the tunnel (dashed line in Fig. 2C). It seems that the bees strive to hold the angular velocity of the image on the wall constant, despite the changes in the angular period of the stripes that accompany the narrowing and widening of the tunnel. This hypothesis is confirmed by the experiments shown in Fig. 3, where bees are made to fly through tunnels of constant width in which the spatial period of the stripes changes abruptly at the halfway point. The bees maintain a constant speed of flight all the way through the tunnel, thus indicating an ability to measure the angular velocity of the image accurately, irrrespective of its spatial structure. Visual control of flight speed is therefore achieved by monitoring and regulating the apparent motion of the visual panorama using a movement-sensitive mechanism that is capable of measuring the angular velocity of the image. A similar conclusion was drawn by David (1982), who investigated visual control of flight speed in the fruitfly Drosophila.

Is it useful to control flight speed by regulating image speed? An obvious advantage of such a strategy is that it would ensure that the bee automatically slows down when negotiating a narrow passage. In addition, it would provide the bee with a simple, safe strategy for landing, as discussed below.

How does a bee execute a smooth touchdown on a horizontal surface? Flight trajectories of landing bees were studied by training bees to collect a reward of sugar water on a horizontal, textured wooden surface, and then removing the reward and video-filming from above the landings that the bees made on the surface in search of the food. The experiments were conducted outdoors on a clear day with the sun at an elevation of approximately 45°. This arrangement allowed the height of the bee to be monitored in terms of the horizontal distance between the bee and its shadow on the wooden surface (Fig. 4A). Height was calibrated in terms of the length of the shadow cast by a vertical rod of a known height (Fig. 4B). This technique, first employed by Zeil (1993), enabled the trajectory of the landing bee to be captured in three dimensions.

A typical trajectory of a landing bee is shown in Fig. 4B. Analysis of such trajectories reveals that the forward speed of the bee decreases steadily as the bee’s height above the surface decreases (Fig. 5). In fact, forward speed is approximately proportional to altitude, indicating that the bee is holding the angular velocity of the image of the surface roughly constant as the surface is approached. This technique may be a simple way of controlling flight speed during landing and ensuring that its value is close to zero at touchdown. The advantage of such a strategy is that control of flight speed is achieved without explicit knowledge of height. Landing bees maintain image angular velocities ranging between 400 and 600 ° s^-1, as revealed by analysis of a number of trajectories.

It is well established that foraging honeybees can gauge the distances travelled to food sources (for a review, see von Frisch, 1967). However, the mechanisms by which they carry out this odometry have remained enigmatic and controversial (Heran, 1956; Neese, 1988; Goller and Esch, 1990; Esch et al. 1994). Very recently, we have investigated whether bees can estimate and learn short distances flown under controlled conditions in the laboratory. Bees were trained to enter a tunnel and collect a reward of sugar solution at a feeder placed in the tunnel at a fixed distance from the entrance (Fig. 6A). The walls and floor of the tunnel were lined with black-and-white stripes perpendicular to the tunnel’s axis. During training, the position and orientation of the tunnel were changed frequently to prevent the bees from using external landmarks to gauge their position in the tunnel. The bees were then tested by recording their searching behaviour in a fresh tunnel which carried no reward and was devoid of any scent cues. Analysis of the test data is described in the legend to Fig. 6B,C. Bees trained in this way showed a clear ability to search for the reward at the correct distance (Fig. 7).

How were the bees gauging the distance flown? Experiments investigating this question yielded the following results. (i) We found that trained bees search at the correct distance from the tunnel entrance even when the entrance aperture to the tunnel is reduced in size or when the length of the tunnel is extended. This behaviour indicates that the bees are not using the visual angle subtended by the entrance or the rear wall to gauge their position in the tunnel (data not shown). (ii) Trained bees search at the correct distance even when the spatial period of the stripes lining the tunnel is changed. Therefore, distance is not gauged by counting the number of stripes or other features passed whilst flying through the tunnel (Fig. 8A). (iii) Trained bees search at the correct distance even when flying against a headwind or with a tailwind, created by a fan placed at the other end of the tunnel (Fig. 8B). In fact, the bees behave as though they overcompensate for the wind, flying slightly farther into the tunnel in a headwind and stopping slightly shorter than the true position of the reward in a tailwind. In a headwind, bees fly more slowly and take longer to reach the estimated location of the reward. The opposite is true in a tailwind. The average times elapsed between entry into the tunnel and the first U-turn are 7.2±2.6 s in still air, 10.1±2.7 s in a headwind of 0.7 m s^-1 and 5.9±1.8 s in a tailwind of 0.65 m s^-1. Therefore, distance is not estimated in terms of time of flight or other correlated parameters, such as number of wingbeats or energy consumption. (iv) When bees trained in a tunnel of a given width are tested in a narrower tunnel, they search at a shorter distance from the entrance; when tested in a wider tunnel, they search at a farther distance (Fig. 8C). This finding suggests that the bees gauge distance flown by integrating the speed of the images of the walls and the floor on the eyes whilst flying through the tunnel. (v) This hypothesis is confirmed in experiments in which image motion is eliminated by using axially oriented stripes on the walls and floor. The bees then display no ability to gauge distance travelled and search uniformly over the entire length of the tunnel (Fig. 8D). The behaviour of the bees in this experiment was strikingly different from that in the other experiments. Here, upon entering the tunnel the bees flew directly to the other end, without turning or even pausing near the former location of the reward. After making a U-turn at the far end, they flew straight back to the entrance. They then exited the tunnel and re-entered it to repeat the above procedure over and over again. Clearly, removal of image-motion cues completely disrupted the bees’ ability to localize the position of the reward. This dramatic loss in performance also argues strongly against the participation of non-visual cues, such as energy consumption, duration of flight or wingbeat counts, in estimating distance flown.

These findings reveal that bees possess a visually driven odometer that measures distance flown by integrating the motion of the image of the visual panorama. They confirm the findings of a recent study that presented indirect evidence supporting a similar conclusion (Esch and Burns, 1995). The results of the experiment of Fig. 8A suggest that, in carrying out this function, the bee’s visual system measures the speed of the image largely independently of the structure of the image.

Ibbotson (1991) reports the existence of visual interneurones in the bee that respond to movement of patterns in the front-to-back direction on each eye. These neurones produce spike frequencies that increase approximately linearly with pattern velocity. Further investigation is needed, however, to establish whether the responses of these neurones are insensitive to pattern structure. If neurones do indeed exist that produce an instantaneous spike frequency proportional to instantaneous image velocity, then the distance travelled along a fixed route could be monitored simply by keeping track of the number of spikes produced.

Our findings do not imply that vision is the sole mechanism by which bees estimate how far they have travelled. The role of other cues, such as energy expenditure, remains controversial (Neese, 1988; Goller and Esch, 1990; Esch et al. 1994; Kirchner and Braun, 1994). Although it is unlikely that energy consumption can be used as a reliable measure of distances as short as those in our experiments (which are traversed in a few seconds), one cannot rule out the importance of such cues in measuring flight over large distances. Unlike energy-based odometry, however, vision-based odometry has the advantage that it is not affected by the presence of wind or by the load of nectar that the bee carries. But visual odometry, as described here, would work accurately only if the bee were to maintain a constant height and follow a fixed route each time it flies to its destination (or if a follower bee follows the same route as a dancing scout bee). This requirement is because the speed of image motion would depend upon the distances to the ground and to the various laterally located objects that are passed on the way to the goal. The consequences of varying these distances are illustrated in the experiments of Fig. 8C.

A recent study suggests that bees may even be able to ‘count’ the number of landmarks passed en route to a feeding station to determine how far they have travelled (Chittka and Geiger, 1995). But this counting does not seem to be the case with the bees flying past stripes, as demonstrated in Fig. 8A. Desert ants foraging in a featureless landscape also seem to use integrated image motion as one cue to estimate distance travelled, although idiothetic cues may play a role as well (Ronacher and Wehner, 1995). Interestingly, the ants’ estimation of distance does not appear to be affected by the burden that they carry (Wehner, 1992).

It is well known that honeybees use celestial as well as landmark-based cues to navigate to a goal. These cues help set the direction of flight, help monitor progress to the goal, provide intermediate beacons at which to aim and aid in pinpointing the target when the bee is in its vicinity (for reviews, see von Frisch, 1967; Wehner, 1992; Collett et al. 1993). In addition, a number of visual cues are used to control flight in a ‘moment-to-moment’ fashion to ensure that the bee stays on course, flies at safe speeds, avoids obstacles and makes a smooth landing at the target. Here we have concentrated on the latter cues. The classically studied optomotor response ensures that the bee stays on course by minimising the apparent rotation of the visual environment (for a review, see Reichardt, 1969). The neural mechanisms subserving this response are well described (for a review, see Hausen and Egelhaaf, 1989). More recent work, described here, suggests that visual control of flight is aided by a number of additional, hitherto unexplored, movement-sensitive mechanisms. Flying bees negotiate narrow passages successfully by balancing the speeds of the images of the flanking walls. Speed of flight is controlled by regulating image velocity, thereby ensuring low speeds whilst passing through narrow gaps. Smooth landings on surfaces are achieved by holding image velocity constant as the surface is approached, thus ensuring low flight speeds at touchdown. Finally, flying bees are able to gauge distance travelled by integrating the optic flow generated by the visual panorama. In all of these tasks, the visual system of the bee displays a capacity to measure the speed of the image independently of the texture of the image. The challenge now is to unravel the neural basis of these navigational feats.