Each artificial flower was equipped with a motion triggered camera that allowed us to record video clips automatically each time a bee was feeding. This way, we could reconstruct the entire flower visitation sequences of bees over several foraging bouts
Dr Mathieu Lihoreau
A study appearing this week in the open access journal PLoS Biology
details how researchers tracked bumblebees to investigate for the first time how they construct the optimal route for visiting multiple flowers and returning to the hive.
The research was carried out by Professor Lars Chittka’s team from the School of Biological and Chemical Sciences at Queen Mary, University of London, working with members of the Harmonic Radar Group at Rothamsted Research. The scientists used radar tracking to monitor the bees’ visits to five artifical flowers in a field 1km in diameter.
questioned co-author Dr Mathieu Lihoreau, now based at the University of Sydney, about the methodology used and the new information on the route optimisation strategies of fauna that can be gleaned from the new research.
While there is a relatively long tradition of research on the foraging behaviour of bees, the methods bees use to localise flowers and establish circuits to revisit them remains very poorly understood. This is mainly because of the challenge posed by accurately following an individual bee from flower to flower over repeated journeys, especially in the field.
"Recently, several research teams, including ours, have started to look at routing behaviour in indoor flight cages using artificial flowers equipped with automated tracking devices to monitor flower visitation sequences," remarked Dr Lihoreau. "These studies showed that bees not only develop regular routes to link multiple flowers but also that these routes are often close to the optimal path – the shortest possible route between flowers."
Based on these discoveries, the team decided to test whether the same type of behaviour can also be observed at larger spatial scales in outdoor conditions. They fitted motion-triggered cameras to five artificial flowers in a 1km diameter field, each possessing a landing platform with a small drop of sucrose at their centre. This enabled the researchers to control the distance between flowers as well as the quantity of nectar provided by each.
"To track the flight paths of bees, we used harmonic radar developed by our colleagues at Rothamsted Research station," explained Dr Lihoreau. "We fixed a transponder to the thorax of the bee. The transponder reflects a component (harmonic) of the waves emitted by a radar positioned in the field. This way we could record the coordinates of the bee in the field in real time when it was exploring the field and moving between flowers.
"Each artificial flower was equipped with a motion triggered camera that allowed us to record video clips automatically each time a bee was feeding. This way, we could reconstruct the entire flower visitation sequences of bees over several foraging bouts."
Mathematical models exposed the basis of the bees’ learning process. Despite initially taking long, complex routes, the bees refined their path through trial and error and discovered the shortest route after trying only 20 of the 120 possibilities. I asked Dr Lihoreau if he and his colleagues were surprised by these results.
"As our recent laboratory studies suggested that bees were able to find the shortest path between multiple flowers, we were hoping that the same would occur in the field," he replied. "However, we didn’t imagine that the optimisation process would be so fast. Clearly, bees have evolved very efficient navigation strategies to minimise their travel distances!"
Dr Lihoreau explained the significance of the findings given that this was the first time route optimisation behaviour could be observed in natural conditions.
"There is a long history of field experiments on bee navigation, beginning with the pioneering work of Nobel Prize winner Karl von Frisch in the 1940s and 50s. But most of this research has focused on how bees learn the location of one flower and develop routes to return to the hive. Of course, everyone knows that bees visit not only one but several hundreds of flowers during a trip. Until now technology simply didn’t allow us to investigate this behaviour in detail."
The ability to observe this behaviour was in itself an innovation, then, but the detailed analysis of the bees’ movements was key to identifying how
bees might proceed to find the shortest possible route.
"Using computer algorithms we were able to simulate the dynamics of route optimisation observed in real bees," said Dr Lihoreau. "Our model suggests that each time a bee tries a novel route, it compares the length of this route to the length of the shortest route experienced so far – we know bees can learn and compare distances. If the new route is found shorter, it is kept in memory, otherwise another solution is tried. Then gradually, through trial and error, the shortest possible can be found."
To test the response of the bees to changes in their environment, the researchers removed one flower and established a new one further away after the bees had been allowed time to learn a stable route between the five artificial flowers.
"The bees that found the new flower immediately included this new flower in their pre-existing route to develop a new optimal solution. This fast response to changes in their environment illustrates just how efficient the optimisation strategy may be in natural environments where flowers disappear and appear on a daily basis."
It is not only larger-brained animals that are capable of learning by trial and error, as this study goes to show. As Dr Lihoreau pointed out, most animals that exploit multiple food sources that are fixed in space and replenish over time face similar route optimisation challenges to the bumblebee.
"This is true for other bee species that collect nectar and pollen from flowers to provision their nest, but also for other nectarivores such as some butterflies, hummingbirds or bats. Some frugivore monkeys as well are known to develop familiar routes for visiting fruit trees. These behaviours are still poorly understood due the difficulty of tracking individual animals in the field, but we would expect all of them to use strategies to minimise their travel distances at some point."