Portal de Periódicos da CAPES

Insect Navigation: Measuring Travel Distance across Ground and through Air

2006; Elsevier BV; Volume: 16; Issue: 20 Linguagem: Inglês

10.1016/j.cub.2006.09.027

1879-0445

M. J. Collett, Thomas S Collett, Mandyam V. Srinivasan,

Insect Pheromone Research and Control

Walking insects probably monitor leg movements to estimate how far they travel, whereas flying insects monitor optic flow. Walking insects probably monitor leg movements to estimate how far they travel, whereas flying insects monitor optic flow. The intriguing story of how honeybees and desert ants measure the long distances that they travel when foraging has unfolded over the last decade or so through numerous studies of their natural behaviour. Bees and ants leave their nest to find food and bring it back for communal use. Desert ants, which do not lay chemical trails, run tens of metres across bare terrain, searching for dead insects and other edible debris. After picking up a food item, they make straight for home and later return directly to the same site in the hope of finding more. Honeybees can also fly directly home from a flower patch several kilometres from the hive, and return later to the same patch. Moreover, by performing waggle dances in the hive, they can communicate the direction and distance of the patch to other foragers so that they too can find it. Navigation of this type involves a strategy known as path integration: an insect monitors the direction and distance of the path that it travels, integrating the two parameters so that it always knows its current direction and distance from the nest. Having found food, the insect stores the path integration coordinates of the site relative to the nest. Path integration thus allows the insect, despite a possibly circuitous outward path, both to travel straight home and also to return straight to its feeding site. In the case of honeybees, it also provides the locational information that is communicated by the waggle dance. Much beautiful research has shown that directional signals used by both insects come from the insect's sky compass. Equally ingenious experiments are now revealing the rather different ways in which bees and ants estimate the distance that they travel. Travel through air and on the ground pose different problems. A major difficulty for a flying insect needing to know its travel distance is that it may be buffeted by head winds or carried by tail winds, so that monitoring motor output or effort cannot give a reliable estimate. Instead, honeybees have been shown to use optic flow to tell them how far they have travelled. As a bee flies above the ground, the world streams across its ventral and lateral retina. This image motion is correlated with travel speed and, through its time integral, with distance travelled. But the use of optic flow has its own complications, of which the most serious is that the speed of retinal image motion depends not only on the bee's ground speed, but also on its distance from objects in its surroundings — primarily height above the ground. For a given travel-distance, the lower a bee flies over a given surface, the greater the flow that it experiences. This characteristic of optic flow may be bothersome to bees, but it provides a good experimental tool for exploring the nature of the bee's odometer, or perhaps more accurately its 'flowometer'. In the mid 1990s, Esch and Burns [1Esch H. Burns J. Distance estimation by foraging honeybees.J. Exp. Biol. 1996; 199: 155-162PubMed Google Scholar] trained bees to take a foraging route between the roofs of two tall buildings. They recorded the waggle dance of returning bees to obtain a measure of how far the bees thought they had flown. The duration of the waggle component, by which the dancing bee tells recruits how far to fly, was shorter than that expressed by bees travelling an equivalent distance closer to the ground. Esch and Burns [1Esch H. Burns J. Distance estimation by foraging honeybees.J. Exp. Biol. 1996; 199: 155-162PubMed Google Scholar] concluded that bees use optic flow to gauge flight distance. Around the same time, Srinivasan et al.[2Srinivasan M.V. Zhang S.W. Lehrer M. Collett T.S. Honeybee navigation en route to the goal: visual flight control and odometry.J. Exp. Biol. 1996; 199: 237-244PubMed Google Scholar] reached the same conclusion by analysing the distances that bees flew through a narrow channel to find food. A channel with textured walls provides powerful optic flow so that a metre flown through a 20 cm wide channel is equivalent to flying a much greater distance over open ground. Bees were trained in a channel of one width to find food at a fixed distance from the channel entrance. In tests, they flew through channels of different widths with no food in the channel. When tested in a channel of the same width as in training, bees flew the expected distance and then searched for the missing food. Bees flown in a narrower channel experienced an increased amount of optic flow per unit of distance travelled relative to the training conditions. Correspondingly, they flew a shorter distance before searching. Bees tested in wider channels experienced less optic flow and flew further before searching. The insects seemed to integrate over time the amount of visual flow and to remember the location of the food in terms of the total amount of flow that they had experienced while flying to the food. With this technique it was also shown that the bees' flowometer is robust to changes in the grain size or pattern of the visual texture on the channel walls [2Srinivasan M.V. Zhang S.W. Lehrer M. Collett T.S. Honeybee navigation en route to the goal: visual flight control and odometry.J. Exp. Biol. 1996; 199: 237-244PubMed Google Scholar, 3Si A. Srinivasan M.V. Zhang S.W. Honeybee navigation: properties of the visually driven 'odometer'.J. Exp. Biol. 2003; 206: 1265-1273Crossref PubMed Scopus (67) Google Scholar] and, above a threshold, to variations in luminance contrast [3Si A. Srinivasan M.V. Zhang S.W. Honeybee navigation: properties of the visually driven 'odometer'.J. Exp. Biol. 2003; 206: 1265-1273Crossref PubMed Scopus (67) Google Scholar]. The bees' estimate of travel distance is also unperturbed by headwinds [2Srinivasan M.V. Zhang S.W. Lehrer M. Collett T.S. Honeybee navigation en route to the goal: visual flight control and odometry.J. Exp. Biol. 1996; 199: 237-244PubMed Google Scholar, 4Srinivasan M.V. Zhang S.W. Bidwell N.J. Visually mediated odometry in honeybees.J. Exp. Biol. 1997; 200: 2513-2522PubMed Google Scholar]. Optic flow signals thus give consistent estimates of flight distance for a variety of environmental conditions and can support path integration, provided that the bee's task is to follow a fixed route and, crucially, that the bee flies at consistent heights along this route. How bees control height is still a mystery. Because bees tend to keep their moment-to-moment optic flow signal constant during flight, increasing their speed as they gain height and reducing it as they approach a surface [2Srinivasan M.V. Zhang S.W. Lehrer M. Collett T.S. Honeybee navigation en route to the goal: visual flight control and odometry.J. Exp. Biol. 1996; 199: 237-244PubMed Google Scholar], optic flow by itself is of little help. Variable winds and loads mean that a combination of thrust and optic flow would also be an unreliable indicator of height. It would be worth testing whether bees employ static cues, such as texture grain size to control height — a bee's cruising height would then vary predictably with the nature of the terrain. Walking insects do not have to rely on translational optic flow and its attendant complications to gauge their speed and distance of travel. Since they are in contact with the ground, unless blown away by a gust of wind, a useful estimate of distance can come from monitoring leg movements or commands. Indeed, while flight speed is strongly influenced by translational optic flow, walking speed in a variety of insects is more or less unaffected by translational flow [5Götz K.G. Wenking H. Visual control of locomotion in the walking fruitfly Drosophila.J. Comp. Physiol. A. 1973; 85: 235-266Crossref Scopus (125) Google Scholar, 6Zanker J.M. Collett T.S. The optomotor system on the ground: on the absence of visual control of speed in walking ladybirds.J. Comp. Physiol. A. 1985; 156: 395-402Crossref Scopus (22) Google Scholar], and recent evidence suggests that the ant's odometer also operates largely independently of translational flow [7Ronacher B. Gallizzi K. Wohlgemuth S. Wehner R. Lateral optic flow does not influence distance estimation in the desert ant Cataglyphis fortis.J. Exp. Biol. 2000; 203: 1113-1121PubMed Google Scholar]. Unlike honeybees, desert ants walking within a narrow channel pay no attention to the width of the channel or to the visual texture on its walls for judging the distance travelled. Furthermore, the ants' estimate of distance is unimpaired when their ventral retina is occluded [7Ronacher B. Gallizzi K. Wohlgemuth S. Wehner R. Lateral optic flow does not influence distance estimation in the desert ant Cataglyphis fortis.J. Exp. Biol. 2000; 203: 1113-1121PubMed Google Scholar]. The ants' odometer can work without visual input from the ventral retina. In a new paper [8Wittlinger M. Wehner R. Wolf H. The ant odometer: stepping on stilts and stumps.Science. 2006; 312: 1965-1967Crossref PubMed Scopus (289) Google Scholar], Wittlinger and colleagues have investigated the problem of desert ant odometry by manipulating the lengths of the ants' legs. In a delicate surgical operation, they either increased leg length by attaching stilts made of pigs' bristles to the ants' legs or they shortened leg length by partial amputation. Ants were trained to run along a straight channel to a feeder. After collecting food, the ants returned home along the same channel. To test their judgement of distance, ants were taken from the feeder and released in a very long test channel (Figure 1A). For normal ants, the length of the return run before the ants started to search for their missing nest, matched the distance of the outward journey. In the experimental groups, the ants' leg-length was manipulated after the ants had reached the feeder. Ants with shortened legs then ran shorter than usual and ants with lengthened legs ran further, as though they had over or under estimated the length of their new stride. Do leg movements play a role in distance measurement? One problem that ants must contend with, when measuring distance, provides a useful clue as to how they might do it. Their speed can be very variable: Ants may walk slowly when first setting out on a foraging trip, run much faster when carrying a biscuit crumb home, or walk very slowly when dragging a large item of prey to their nest. They adjust their speed by changing both stride length and stride frequency [9Zollikofer C.P.E. Stepping patterns in ants. I. Influence of speed and curvature.J. Exp. Biol. 1994; 192: 95-106PubMed Google Scholar] (Figure 2). Thus, ants cannot gauge distance accurately simply by counting steps. Each stride counted must be tagged with a length. Measurements from high-speed video recordings of the operated ants (Table 1) support the conclusion that ants do something more than step-counting. The ants would have taken on average 770 steps to travel the 10 m to the food, but while the normal ants also took 770 steps on their 10 m return, the ants whose legs were then shortened searched for their nest at 5.75 m after 670 steps, while the ants with stilts added began searching at 15.3 m after having taken 1030 steps. If the three groups perceive these distances to be the same, then paradoxically the ants with the shortened stumps must suppose their strides to be longer than normal while the ants on stilts must perceive their strides to be shorter!Table 1Effects of operation on walking.TreatmentDistance travelledStride-lengthSpeedNumber of steps∗Perceived stride-length∗Stride-frequency∗Shortened5.75 m8.6 mm140 mm/s67014.9 mm16 HzNormal10 m13 mm310 mm/s77013 mm24 HzLengthened15.3 m14.8 mm290 mm/s10309.7 mm20 HzThe first three columns give data from 8Wittlinger M. Wehner R. Wolf H. The ant odometer: stepping on stilts and stumps.Science. 2006; 312: 1965-1967Crossref PubMed Scopus (289) Google Scholar. Columns marked with an askerisk give our calculations from the data. Number of steps = (Distance travelled)/(Stride-length); perceived stride-length = (Normal stride-length)∗(Normal number of steps)/(Number of steps); stride frequency = (Speed)/(Stride-length). Open table in a new tab The first three columns give data from 8Wittlinger M. Wehner R. Wolf H. The ant odometer: stepping on stilts and stumps.Science. 2006; 312: 1965-1967Crossref PubMed Scopus (289) Google Scholar. Columns marked with an askerisk give our calculations from the data. Number of steps = (Distance travelled)/(Stride-length); perceived stride-length = (Normal stride-length)∗(Normal number of steps)/(Number of steps); stride frequency = (Speed)/(Stride-length). So why should perceived stride-length change in the observed manner? Stride-length tends to increase monotonically with stride-frequency (Figure 2). But while the ants with stumps have a longer perceived stride-length (14.9 mm) than the ants on stilts (perceived stride-length 9.7 mm), their stride-frequency is lower (16 Hz versus 20 Hz). The operations must have an additional effect on perceived stride-length that is independent of stride-frequency. We offer a possible explanation that stride-length is measured through proprioception of the joint-angles of the legs. Because stumps weigh less than whole legs, they would be swung more, so exaggerating perceived stride-length. Ants on stumps would therefore take fewer steps home, believing each step to be longer than it is. In contrast, the stilts are heavy and are swung less, so that stride length appears to be shorter than normal. Thus, ants on stilts would take more steps before believing themselves to have arrived home. Wittlinger et al.'s [8Wittlinger M. Wehner R. Wolf H. The ant odometer: stepping on stilts and stumps.Science. 2006; 312: 1965-1967Crossref PubMed Scopus (289) Google Scholar] data suggest therefore that Cataglyphis estimates its journey distance by summing total stride length over the whole journey and that stride length may be estimated using proprioceptive inputs. It might be interesting to examine the video recordings to see whether the swing in each stride is indeed greater with stumps than with stilts. A slight caveat to interpreting these experiments is that attaching stilts raises an ant above the ground whereas amputating its legs lowers it towards the ground, so that similar effects of these operations on distance estimation could be predicted by the use of an odometer driven by ventral optic flow. Thus, the correctness of Wittlinger et al.'s conclusions relies heavily on the earlier studies, which suggest that ventral optic flow plays at most a very minor part in the ant's odometer [7Ronacher B. Gallizzi K. Wohlgemuth S. Wehner R. Lateral optic flow does not influence distance estimation in the desert ant Cataglyphis fortis.J. Exp. Biol. 2000; 203: 1113-1121PubMed Google Scholar]. Mindful of this problem, Wittlinger et al.[8Wittlinger M. Wehner R. Wolf H. The ant odometer: stepping on stilts and stumps.Science. 2006; 312: 1965-1967Crossref PubMed Scopus (289) Google Scholar] tested ants on a smooth surface of fine sand hoping to eliminate any cues from optic flow. The small size of ants in relation to the microstructure of the ground and their frequent deviations from a straight path, both horizontally and vertically, make distance measurements of little use without accompanying directional information. In line with most models of path integration, one would expect the odometer to incorporate directional information when recording stride-length. Two recent experimental studies emphasise this point. First, Sommer and Wehner [10Sommer S. Wehner R. Vector navigation in desert ants, Cataglyphis fortis: celestial compass cues are essential for the proper use of distance information.Naturwiss. 2005; 92: 468-471Crossref PubMed Scopus (28) Google Scholar] found that if desert ants run part of their route with no input from the sky compass, that portion of the route is unrecorded. Ants ran between their nest and a feeder, either along an open topped channel that gave a continuous view of the sky or with the channel divided into 1.5 m segments, with segments alternately open to the sky or covered with wood, so depriving the ants of compass cues (Figure 1B). Ants taken from the feeder were placed in a long open-topped channel. The experimental ants trained in the partially covered channel ran half as far as ants trained in the uncovered channel. Distance is registered only when there is a concurrent directional signal. Ants without sky compass cues take no risks and make no assumptions about their direction of travel. The second study by Wohlgemuth et al.[11Wohlgemuth S. Ronacher B. Wehner R. Distance estimation in the third dimension in desert ants.J. Comp. Physiol. A. 2002; 188: 273-281Crossref Scopus (42) Google Scholar] shows that when ants walk over undulating terrain, they do not measure the total distance travelled, including the undulations. Rather, they seem to register only the projected distance travelled in the horizontal plane. If an ant is made to walk to the feeder on a roller coaster and return home on the flat, the distance walked on the return excludes the vertical component of the roller coaster path (Figure 1C). As yet it is unclear whether ants are performing true path integration in three dimension, as do jumping spiders [12Hill D.E. Orientation by jumping spiders of the genus Phidippus (Araneae: Salticidae) during the pursuit of prey.Behav. Ecol. Sociobiol. 1979; 6: 301-322Crossref Scopus (70) Google Scholar], or whether they are merely discounting the vertical component of their path. But what all these studies do make abundantly clear is that both the ants' and the bees' navigational strategies have evolved to match closely the ecological constraints of their particular environments.